diff options
Diffstat (limited to 'extras')
80 files changed, 7895 insertions, 753 deletions
diff --git a/extras/FreeBSD/Makefile b/extras/FreeBSD/Makefile index c662b9c65..26fe85c88 100644 --- a/extras/FreeBSD/Makefile +++ b/extras/FreeBSD/Makefile @@ -10,17 +10,17 @@ DISTVERSION= 2.0.0rc1 PORTVERSION= 2.0.0rc1 PORTREVISION= 1 CATEGORIES= sysutils -MASTER_SITES= ftp://ftp.zresearch.com/pub/gluster/glusterfs/2.0/2.0.0/ \ +MASTER_SITES= ftp://ftp.gluster.com/pub/gluster/glusterfs/2.0/2.0.0/ \ http://europe.gluster.org/glusterfs/2.0/2.0.0/ -MAINTAINER= harsha@zresearch.com +MAINTAINER= harsha@gluster.com COMMENT= GlusterFS is a clustered file-system BUILD_DEPENDS= fusefs-libs>2.6.3:${PORTSDIR}/sysutils/fusefs-libs LIB_DEPENDS= fuse.2:${PORTSDIR}/sysutils/fusefs-libs RUN_DEPENDS= ${LOCALBASE}/modules/fuse.ko:${PORTSDIR}/sysutils/fusefs-kmod -MAN8= glusterfs.8 +MAN8= glusterfs.8 gluster.8 glusterd.8 glusterfs-volgen.8 GNU_CONFIGURE= yes USE_LDCONFIG= yes USE_AUTOTOOLS= libtool:15 diff --git a/extras/LinuxRPM/Makefile.am b/extras/LinuxRPM/Makefile.am new file mode 100644 index 000000000..1dafa982b --- /dev/null +++ b/extras/LinuxRPM/Makefile.am @@ -0,0 +1,57 @@ + +GFS_TAR = ../../glusterfs-$(VERSION).tar.gz + +.PHONY: all + +all: + @echo "To build RPMS run 'make glusterrpms'" + +.PHONY: glusterrpms prep srcrpm testsrpm clean + +glusterrpms: prep srcrpm rpms + -rm -rf rpmbuild + +prep: + if [ ! -e $(GFS_TAR) ]; then \ + $(MAKE) -C ../.. dist; \ + fi + -mkdir -p rpmbuild/SPECS + -mkdir -p rpmbuild/RPMS + -mkdir -p rpmbuild/SRPMS + -rm -rf rpmbuild/SOURCES + @if [ -d /d/cache/glusterfs -a -e /d/cache/glusterfs/sources ]; then \ + echo "copying glusterfs rpm files from local cache..." ; \ + mkdir -p ./rpmbuild/SOURCES; \ + cp /d/cache/glusterfs/* ./rpmbuild/SOURCES/ ; \ + elif [ -x /usr/bin/git ]; then \ + echo "fetching glusterfs rpm files from fedora git repo..."; \ + cd ./rpmbuild && git clone git://pkgs.fedoraproject.org/glusterfs.git > /dev/null && mv glusterfs SOURCES; \ + else \ + echo "glusterfs rpm files not fetched, you don't have git installed!" ; \ + exit 1 ; \ + fi + cp ../../*.tar.gz ./rpmbuild/SOURCES + cp ../../glusterfs.spec ./rpmbuild/SPECS + +srcrpm: + rpmbuild --define '_topdir $(shell pwd)/rpmbuild' -bs rpmbuild/SPECS/glusterfs.spec + mv rpmbuild/SRPMS/* . + +rpms: + rpmbuild --define '_topdir $(shell pwd)/rpmbuild' -bb rpmbuild/SPECS/glusterfs.spec + mv rpmbuild/RPMS/*/* . + +# EPEL-5 does not like new versions of rpmbuild and requires some +# _source_* defines + +testsrpm: prep + rpmbuild --define '_topdir $(shell pwd)/rpmbuild' \ + --define '_source_payload w9.gzdio' \ + --define '_source_filedigest_algorithm 1' \ + -bs rpmbuild/SPECS/glusterfs.spec + mv rpmbuild/SRPMS/* ../.. + -rm -rf rpmbuild + +clean: + -rm -rf rpmbuild + -rm -f *.rpm diff --git a/extras/MacOSX/Portfile b/extras/MacOSX/Portfile index 0094ee4c4..1262a44da 100644 --- a/extras/MacOSX/Portfile +++ b/extras/MacOSX/Portfile @@ -5,12 +5,12 @@ PortSystem 1.0 name glusterfs version 2.0.0rc8 categories fuse -maintainers amar@zresearch.com +maintainers amar@gluster.com description GlusterFS long_description GlusterFS is a cluster file system, flexible to tune it for your needs. homepage http://www.gluster.org/ platforms darwin -master_sites http://ftp.zresearch.com/pub/gluster/glusterfs/2.0/2.0.0 +master_sites http://ftp.gluster.com/pub/gluster/glusterfs/2.0/2.0.0 configure.args --disable-bdb checksums md5 33c2d02344d4fab422e80cfb637e0b48 @@ -18,7 +18,7 @@ checksums md5 33c2d02344d4fab422e80cfb637e0b48 post-destroot { file mkdir ${destroot}/Library/LaunchDaemons/ file copy ${worksrcpath}/extras/glusterfs-server.plist \ - ${destroot}/Library/LaunchDaemons/com.zresearch.glusterfs.plist + ${destroot}/Library/LaunchDaemons/com.gluster.glusterfs.plist file mkdir ${destroot}/sbin/ file copy ${worksrcpath}/xlators/mount/fuse/utils/mount_glusterfs \ diff --git a/extras/MacOSX/README.MacOSX b/extras/MacOSX/README.MacOSX index cc38958d8..ec7abc2cc 100644 --- a/extras/MacOSX/README.MacOSX +++ b/extras/MacOSX/README.MacOSX @@ -24,7 +24,7 @@ system. Run, bash# sudo /Library/Filesystems/fusefs.fs/Support/uninstall-macfuse-core.sh After this, install MacFUSE (mostly through .dmg available in macfuse homepage -or if Z Research Inc provides any custom built .dmg) +or if Gluster Inc provides any custom built .dmg) Make sure the .dmg of glusterfs is built against the installed MacFUSE version (if not, any operations over mountpoint gives EIO ie, Input/Output Error). If @@ -40,7 +40,7 @@ after the entry is added in /etc/fstab, it can be mounted by 'mount' command. To start the server process one can use the 'launchd' mechanism. Follow below steps after installation - bash# launchctl load /Library/LaunchDaemons/com.zresearch.glusterfs.plist + bash# launchctl load /Library/LaunchDaemons/com.gluster.glusterfs.plist No need to run the command if the machine reboots, as it will be loaded automatically by launchd process. @@ -51,17 +51,17 @@ Now copy the server volume file in the proper path NOTE: (If glusterfs is installed in different path other than '/opt/local' update the volume file at the corresponding path, and also need to - update the /Library/LaunchDaemons/com.zresearch.glusterfs.plist with + update the /Library/LaunchDaemons/com.gluster.glusterfs.plist with the proper path) Once the volume file is updated, administrator can start the server process by running, - bash# launchctl start com.zresearch.glusterfs + bash# launchctl start com.gluster.glusterfs and stop like - bash# launchctl stop com.zresearch.glusterfs + bash# launchctl stop com.gluster.glusterfs NOTE: To start the process by default when the loaded, one need to add the following lines to .plist file diff --git a/extras/Makefile.am b/extras/Makefile.am index 9be397300..cf619329b 100644 --- a/extras/Makefile.am +++ b/extras/Makefile.am @@ -1,11 +1,20 @@ -docdir = $(datadir)/doc/glusterfs/ -EditorModedir = $(docdir)/ +EditorModedir = $(docdir) EditorMode_DATA = glusterfs-mode.el glusterfs.vim -SUBDIRS = init.d benchmarking +SUBDIRS = init.d systemd benchmarking hook-scripts $(OCF_SUBDIR) LinuxRPM geo-rep -EXTRA_DIST = specgen.scm MacOSX/Portfile glusterfs-mode.el glusterfs.vim migrate-unify-to-distribute.sh backend-xattr-sanitize.sh +confdir = $(sysconfdir)/glusterfs +conf_DATA = glusterfs-logrotate gluster-rsyslog-7.2.conf gluster-rsyslog-5.8.conf \ + logger.conf.example glusterfs-georep-logrotate -CLEANFILES = +voldir = $(sysconfdir)/glusterfs +vol_DATA = glusterd.vol +EXTRA_DIST = specgen.scm MacOSX/Portfile glusterfs-mode.el glusterfs.vim \ + migrate-unify-to-distribute.sh backend-xattr-sanitize.sh \ + backend-cleanup.sh disk_usage_sync.sh quota-remove-xattr.sh \ + quota-metadata-cleanup.sh glusterfs-logrotate clear_xattrs.sh \ + group-virt.example glusterd-sysconfig gluster-rsyslog-7.2.conf \ + gluster-rsyslog-5.8.conf logger.conf.example glusterd.vol \ + glusterfs-georep-logrotate diff --git a/extras/Solaris/Prototype b/extras/Solaris/Prototype index 369234210..cb2d99d8c 100644 --- a/extras/Solaris/Prototype +++ b/extras/Solaris/Prototype @@ -9,6 +9,9 @@ d none /usr/sfw/share 0755 root bin d none /usr/sfw/share/man 0755 root bin d none /usr/sfw/share/man/man8 0755 root bin f none /usr/sfw/share/man/man8/glusterfs.8 0644 root root +f none /usr/sfw/share/man/man8/gluster.8 0644 root root +f none /usr/sfw/share/man/man8/glusterd.8 0644 root root +f none /usr/sfw/share/man/man8/glusterfs-volgen.8 0644 root root d none /usr/sfw/share/doc 0755 root bin d none /usr/sfw/share/doc/glusterfs 0755 root root d none /usr/sfw/share/doc/glusterfs/examples 0755 root root diff --git a/extras/Solaris/README.solaris b/extras/Solaris/README.solaris index 0cea45bdc..5a4e4336d 100644 --- a/extras/Solaris/README.solaris +++ b/extras/Solaris/README.solaris @@ -10,7 +10,7 @@ $ which glusterfs $ glusterfs --version glusterfs 2.0.0rc1 built on Jan 16 2009 03:36:59 Repository revision: glusterfs--mainline--3.0--patch-844 -Copyright (c) 2006, 2007, 2008 Z RESEARCH Inc. <http://www.zresearch.com> +Copyright (c) 2006-2011 Gluster Inc. <http://www.gluster.com> GlusterFS comes with ABSOLUTELY NO WARRANTY. You may redistribute copies of GlusterFS under the terms of the GNU General Public License. diff --git a/extras/Solaris/pkginfo b/extras/Solaris/pkginfo index ffa1f2d9a..0da4907d8 100644 --- a/extras/Solaris/pkginfo +++ b/extras/Solaris/pkginfo @@ -4,9 +4,9 @@ VERSION="2.0.0rc1" ARCH="i386" CLASSES="none" CATEGORY="system" -VENDOR="ZResearch Inc" +VENDOR="Gluster Inc" PSTAMP="16thJan08" -EMAIL="harsha@zresearch.com" +EMAIL="harsha@gluster.com" ISTATES="S s 1 2 3" RSTATES="S s 1 2 3" BASEDIR="/" diff --git a/extras/Ubuntu/README.Ubuntu b/extras/Ubuntu/README.Ubuntu new file mode 100644 index 000000000..0c5b7828d --- /dev/null +++ b/extras/Ubuntu/README.Ubuntu @@ -0,0 +1,24 @@ +Bug 765014 - Mounting from localhost in fstab fails at boot on ubuntu +(https://bugzilla.redhat.com/show_bug.cgi?id=765014) +(https://bugs.launchpad.net/ubuntu/+source/glusterfs/+bug/876648) + +Ubuntu uses upstart instead of init to bootstrap the system and it has a unique +way of handling fstab, using a program called mountall(8). As a result of using +a debian initscript to start glusterd, glusterfs mounts in fstab are tried before +the glusterd service is running. In the case where the client is also a server +and the volume is mounted from localhost, the mount fails at boot time. To +correct this we need to launch glusterd using upstart and block the glusterfs +mounting event until glusterd is started. + +The glusterd.conf file contains the necessary configuration for upstart to +manage the glusterd service. It should be placed in /etc/init/glusterd.conf +on Ubuntu systems, and then the old initscript /etc/init.d/glusterd can be +removed. An additional upstart job, mounting-glusterfs.conf, is also required +to block mounting glusterfs volumes until the glusterd service is available. +Both of these upstart jobs need to be placed in /etc/init to resolve the issue. + +Starting with Ubuntu 12.04, Precise Pangolin, these upstart jobs will be +included with the glusterfs-server package in the Ubuntu repository. + +This affects all versions of glusterfs on the Ubuntu platform since at least +10.04, Lucid Lynx. diff --git a/extras/Ubuntu/glusterd.conf b/extras/Ubuntu/glusterd.conf new file mode 100644 index 000000000..aa99502b0 --- /dev/null +++ b/extras/Ubuntu/glusterd.conf @@ -0,0 +1,10 @@ +author "Louis Zuckerman <me@louiszuckerman.com>" +description "GlusterFS Management Daemon" + +start on runlevel [2345] +stop on runlevel [016] + +expect fork + +exec /usr/sbin/glusterd -p /var/run/glusterd.pid + diff --git a/extras/Ubuntu/mounting-glusterfs.conf b/extras/Ubuntu/mounting-glusterfs.conf new file mode 100644 index 000000000..3c59c0f63 --- /dev/null +++ b/extras/Ubuntu/mounting-glusterfs.conf @@ -0,0 +1,7 @@ +author "Louis Zuckerman <me@louiszuckerman.com>" +description "Block the mounting event for glusterfs filesystems until glusterd is running" + +start on mounting TYPE=glusterfs +task +exec start wait-for-state WAIT_FOR=glusterd WAITER=mounting-glusterfs + diff --git a/extras/backend-cleanup.sh b/extras/backend-cleanup.sh new file mode 100644 index 000000000..dc504860b --- /dev/null +++ b/extras/backend-cleanup.sh @@ -0,0 +1,28 @@ +#!/bin/sh + +# This script can be used to cleanup the 'cluster/distribute' translator's +# stale link files. One may choose to run this only when number of subvolumes +# to distribute volume gets increased (or decreased) +# +# This script has to be run on the servers, which are exporting the data to +# GlusterFS +# +# (c) 2010 Gluster Inc <http://www.gluster.com/> + +set -e + +# Change the below variable as per the setup. +export_directory="/export/glusterfs" + +clean_dir() +{ + # Clean the 'link' files on backend + find "${export_directory}" -type f -perm +01000 -exec rm -v '{}' \; +} + +main() +{ + clean_dir ; +} + +main "$@" diff --git a/extras/benchmarking/Makefile.am b/extras/benchmarking/Makefile.am index 04cc06182..bfcc59277 100644 --- a/extras/benchmarking/Makefile.am +++ b/extras/benchmarking/Makefile.am @@ -1,7 +1,5 @@ -docdir = $(datadir)/doc/$(PACKAGE_NAME)/benchmarking - -benchmarkingdir = $(docdir) +benchmarkingdir = $(docdir)/benchmarking benchmarking_DATA = rdd.c glfs-bm.c README launch-script.sh local-script.sh diff --git a/extras/benchmarking/glfs-bm.c b/extras/benchmarking/glfs-bm.c index 66b32b3f0..dc717f33c 100644 --- a/extras/benchmarking/glfs-bm.c +++ b/extras/benchmarking/glfs-bm.c @@ -1,28 +1,18 @@ -/* - Copyright (c) 2008-2009 Z RESEARCH, Inc. <http://www.zresearch.com> - This file is part of GlusterFS. - GlusterFS is free software; you can redistribute it and/or modify - it under the terms of the GNU General Public License as published - by the Free Software Foundation; either version 3 of the License, - or (at your option) any later version. - - GlusterFS is distributed in the hope that it will be useful, but - WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - General Public License for more details. - - You should have received a copy of the GNU General Public License - along with this program. If not, see - <http://www.gnu.org/licenses/>. +/* + Copyright (c) 2008-2012 Red Hat, Inc. <http://www.redhat.com> + This file is part of GlusterFS. + + This file is licensed to you under your choice of the GNU Lesser + General Public License, version 3 or any later version (LGPLv3 or + later), or the GNU General Public License, version 2 (GPLv2), in all + cases as published by the Free Software Foundation. */ - #define _GNU_SOURCE #define __USE_FILE_OFFSET64 #define _FILE_OFFSET_BITS 64 #include <stdio.h> #include <argp.h> -#include <libglusterfsclient.h> #include <stdlib.h> #include <unistd.h> #include <sys/types.h> @@ -42,7 +32,6 @@ struct state { char need_iface_xattr:1; char need_mode_posix:1; - char need_mode_libglusterfsclient:1; char prefix[512]; long int count; @@ -50,7 +39,6 @@ struct state { size_t block_size; char *specfile; - void *libglusterfsclient_context; long int io_size; }; @@ -132,21 +120,6 @@ parse_opts (int key, char *arg, return -1; } break; - case 'm': - if (strcasecmp (arg, "posix") == 0) { - state->need_mode_posix = 1; - state->need_mode_libglusterfsclient = 0; - } else if (strcasecmp (arg, "libglusterfsclient") == 0) { - state->need_mode_posix = 0; - state->need_mode_libglusterfsclient = 1; - } else if (strcasecmp (arg, "both") == 0) { - state->need_mode_posix = 1; - state->need_mode_libglusterfsclient = 1; - } else { - fprintf (stderr, "unknown mode: %s\n", arg); - return -1; - } - break; case 'b': { size_t block_size = atoi (arg); @@ -342,170 +315,6 @@ do_mode_posix_iface_xattr (struct state *state) return 0; } - -int -do_mode_libglusterfsclient_iface_fileio_write (struct state *state) -{ - long int i; - int ret = -1; - char block[state->block_size]; - - for (i=0; i<state->count; i++) { - long fd = 0; - char filename[512]; - - sprintf (filename, "/%s.%06ld", state->prefix, i); - - fd = glusterfs_open (state->libglusterfsclient_context, - filename, O_CREAT|O_WRONLY, 0); - - if (fd == 0) { - fprintf (stderr, "open(%s) => %s\n", filename, strerror (errno)); - break; - } - ret = glusterfs_write (fd, block, state->block_size); - if (ret == -1) { - fprintf (stderr, "glusterfs_write(%s) => %s\n", filename, strerror (errno)); - glusterfs_close (fd); - break; - } - glusterfs_close (fd); - state->io_size += ret; - } - - return i; -} - - -int -do_mode_libglusterfsclient_iface_fileio_read (struct state *state) -{ - long int i; - int ret = -1; - char block[state->block_size]; - - for (i=0; i<state->count; i++) { - long fd = 0; - char filename[512]; - - sprintf (filename, "/%s.%06ld", state->prefix, i); - - fd = glusterfs_open (state->libglusterfsclient_context, - filename, O_RDONLY, 0); - - if (fd == 0) { - fprintf (stderr, "glusterfs_open(%s) => %s\n", filename, strerror (errno)); - break; - } - ret = glusterfs_read (fd, block, state->block_size); - if (ret == -1) { - fprintf (stderr, "glusterfs_read(%s) => %s\n", filename, strerror (errno)); - glusterfs_close (fd); - break; - } - glusterfs_close (fd); - state->io_size += ret; - } - - return i; -} - - -int -do_mode_libglusterfsclient_iface_fileio (struct state *state) -{ - if (state->need_op_write) - MEASURE (do_mode_libglusterfsclient_iface_fileio_write, state); - - if (state->need_op_read) - MEASURE (do_mode_libglusterfsclient_iface_fileio_read, state); - - return 0; -} - - -int -do_mode_libglusterfsclient_iface_xattr_write (struct state *state) -{ - long int i; - int ret = -1; - char block[state->block_size]; - char *dname = NULL, *dirc = NULL; - char *bname = NULL, *basec = NULL; - - asprintf (&dirc, "/%s", state->prefix); - asprintf (&basec, "/%s", state->prefix); - dname = dirname (dirc); - bname = basename (basec); - - for (i=0; i<state->count; i++) { - char key[512]; - - sprintf (key, "glusterfs.file.%s.%06ld", bname, i); - - ret = glusterfs_setxattr (state->libglusterfsclient_context, - dname, key, block, state->block_size, 0); - - if (ret < 0) { - fprintf (stderr, "glusterfs_setxattr (%s, %s, %p) => %s\n", - dname, key, block, strerror (errno)); - break; - } - state->io_size += state->block_size; - } - - return i; - -} - - -int -do_mode_libglusterfsclient_iface_xattr_read (struct state *state) -{ - long int i; - int ret = -1; - char block[state->block_size]; - char *dname = NULL, *dirc = NULL; - char *bname = NULL, *basec = NULL; - - dirc = strdup (state->prefix); - basec = strdup (state->prefix); - dname = dirname (dirc); - bname = basename (basec); - - for (i=0; i<state->count; i++) { - char key[512]; - - sprintf (key, "glusterfs.file.%s.%06ld", bname, i); - - ret = glusterfs_getxattr (state->libglusterfsclient_context, - dname, key, block, state->block_size); - - if (ret < 0) { - fprintf (stderr, "glusterfs_getxattr (%s, %s, %p) => %s\n", - dname, key, block, strerror (errno)); - break; - } - state->io_size += ret; - } - - return i; -} - - -int -do_mode_libglusterfsclient_iface_xattr (struct state *state) -{ - if (state->need_op_write) - MEASURE (do_mode_libglusterfsclient_iface_xattr_write, state); - - if (state->need_op_read) - MEASURE (do_mode_libglusterfsclient_iface_xattr_read, state); - - return 0; -} - - int do_mode_posix (struct state *state) { @@ -520,44 +329,8 @@ do_mode_posix (struct state *state) int -do_mode_libglusterfsclient (struct state *state) -{ - glusterfs_init_ctx_t ctx = { - .logfile = "/dev/stderr", - .loglevel = "error", - .lookup_timeout = 60, - .stat_timeout = 60, - }; - - ctx.specfile = state->specfile; - if (state->specfile) { - state->libglusterfsclient_context = glusterfs_init (&ctx); - - if (!state->libglusterfsclient_context) { - fprintf (stdout, "Unable to initialize glusterfs context, skipping libglusterfsclient mode\n"); - return -1; - } - } else { - fprintf (stdout, "glusterfs volume specification file not provided, skipping libglusterfsclient mode\n"); - return -1; - } - - if (state->need_iface_fileio) - do_mode_libglusterfsclient_iface_fileio (state); - - if (state->need_iface_xattr) - do_mode_libglusterfsclient_iface_xattr (state); - - return 0; -} - - -int do_actions (struct state *state) { - if (state->need_mode_libglusterfsclient) - do_mode_libglusterfsclient (state); - if (state->need_mode_posix) do_mode_posix (state); @@ -569,8 +342,6 @@ static struct argp_option options[] = { "WRITE|READ|BOTH - defaults to BOTH"}, {"iface", 'i', "INTERFACE", 0, "FILEIO|XATTR|BOTH - defaults to FILEIO"}, - {"mode", 'm', "MODE", 0, - "POSIX|LIBGLUSTERFSCLIENT|BOTH - defaults to POSIX"}, {"block", 'b', "BLOCKSIZE", 0, "<NUM> - defaults to 4096"}, {"specfile", 's', "SPECFILE", 0, @@ -601,7 +372,6 @@ main (int argc, char *argv[]) state.need_iface_xattr = 0; state.need_mode_posix = 1; - state.need_mode_libglusterfsclient = 0; state.block_size = 4096; diff --git a/extras/benchmarking/rdd.c b/extras/benchmarking/rdd.c index d30660b50..a667c6a1d 100644 --- a/extras/benchmarking/rdd.c +++ b/extras/benchmarking/rdd.c @@ -1,22 +1,12 @@ /* - Copyright (c) 2008 Z RESEARCH, Inc. <http://www.zresearch.com> - This file is part of GlusterFS. - - GlusterFS is free software; you can redistribute it and/or modify - it under the terms of the GNU General Public License as published - by the Free Software Foundation; either version 3 of the License, - or (at your option) any later version. - - GlusterFS is distributed in the hope that it will be useful, but - WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - General Public License for more details. - - You should have received a copy of the GNU General Public License - along with this program. If not, see - <http://www.gnu.org/licenses/>. -*/ + Copyright (c) 2008-2012 Red Hat, Inc. <http://www.redhat.com> + This file is part of GlusterFS. + This file is licensed to you under your choice of the GNU Lesser + General Public License, version 3 or any later version (LGPLv3 or + later), or the GNU General Public License, version 2 (GPLv2), in all + cases as published by the Free Software Foundation. +*/ #include <stdio.h> #include <sys/stat.h> #include <sys/types.h> @@ -36,172 +26,278 @@ #define UNIX_PATH_MAX 108 #endif +#define UNIT_KB 1024ULL +#define UNIT_MB UNIT_KB*1024ULL +#define UNIT_GB UNIT_MB*1024ULL +#define UNIT_TB UNIT_GB*1024ULL +#define UNIT_PB UNIT_TB*1024ULL + +#define UNIT_KB_STRING "KB" +#define UNIT_MB_STRING "MB" +#define UNIT_GB_STRING "GB" +#define UNIT_TB_STRING "TB" +#define UNIT_PB_STRING "PB" + struct rdd_file { - char path[UNIX_PATH_MAX]; - struct stat st; - int fd; + char path[UNIX_PATH_MAX]; + struct stat st; + int fd; }; - + struct rdd_config { - long iters; - long max_ops_per_seq; - size_t max_bs; - size_t min_bs; - int thread_count; - pthread_t *threads; - pthread_barrier_t barrier; - pthread_mutex_t lock; - struct rdd_file in_file; - struct rdd_file out_file; + long iters; + long max_ops_per_seq; + size_t max_bs; + size_t min_bs; + int thread_count; + pthread_t *threads; + pthread_barrier_t barrier; + pthread_mutex_t lock; + struct rdd_file in_file; + struct rdd_file out_file; + ssize_t file_size; }; static struct rdd_config rdd_config; enum rdd_keys { - RDD_MIN_BS_KEY = 1, - RDD_MAX_BS_KEY, + RDD_MIN_BS_KEY = 1, + RDD_MAX_BS_KEY, }; - + static error_t rdd_parse_opts (int key, char *arg, - struct argp_state *_state) + struct argp_state *_state) { - switch (key) { - case 'o': - { - int len = 0; - len = strlen (arg); - if (len > UNIX_PATH_MAX) { - fprintf (stderr, "output file name too long (%s)\n", + switch (key) { + case 'o': + { + int len = 0; + len = strlen (arg); + if (len > UNIX_PATH_MAX) { + fprintf (stderr, "output file name too long (%s)\n", arg); - return -1; - } - - strncpy (rdd_config.out_file.path, arg, len); - } - break; - - case 'i': - { - int len = 0; - len = strlen (arg); - if (len > UNIX_PATH_MAX) { - fprintf (stderr, "input file name too long (%s)\n", + return -1; + } + + strncpy (rdd_config.out_file.path, arg, len); + } + break; + + case 'i': + { + int len = 0; + len = strlen (arg); + if (len > UNIX_PATH_MAX) { + fprintf (stderr, "input file name too long (%s)\n", arg); - return -1; - } - - strncpy (rdd_config.in_file.path, arg, len); - } - break; - - case RDD_MIN_BS_KEY: - { - char *tmp = NULL; - long bs = 0; - bs = strtol (arg, &tmp, 10); - if ((bs == LONG_MAX) || (bs == LONG_MIN) || (tmp && *tmp)) { - fprintf (stderr, "invalid argument for minimum block" + return -1; + } + + strncpy (rdd_config.in_file.path, arg, len); + rdd_config.in_file.path[len] = '\0'; + } + break; + + case 'f': + { + char *tmp = NULL; + unsigned long long fs = 0; + if (string2bytesize (arg, &fs) == -1) { + fprintf (stderr, "invalid argument for file size " + "(%s)\n", arg); + return -1; + } + + rdd_config.file_size = fs; + } + break; + + case RDD_MIN_BS_KEY: + { + char *tmp = NULL; + long bs = 0; + bs = strtol (arg, &tmp, 10); + if ((bs == LONG_MAX) || (bs == LONG_MIN) || (tmp && *tmp)) { + fprintf (stderr, "invalid argument for minimum block" "size (%s)\n", arg); - return -1; - } - - rdd_config.min_bs = bs; - } - break; - - case RDD_MAX_BS_KEY: - { - char *tmp = NULL; - long bs = 0; - bs = strtol (arg, &tmp, 10); - if ((bs == LONG_MAX) || (bs == LONG_MIN) || (tmp && *tmp)) { - fprintf (stderr, "invalid argument for maximum block" + return -1; + } + + rdd_config.min_bs = bs; + } + break; + + case RDD_MAX_BS_KEY: + { + char *tmp = NULL; + long bs = 0; + bs = strtol (arg, &tmp, 10); + if ((bs == LONG_MAX) || (bs == LONG_MIN) || (tmp && *tmp)) { + fprintf (stderr, "invalid argument for maximum block" "size (%s)\n", arg); - return -1; - } - - rdd_config.max_bs = bs; - } - break; - - case 'r': - { - char *tmp = NULL; - long iters = 0; - iters = strtol (arg, &tmp, 10); - if ((iters == LONG_MAX) || - (iters == LONG_MIN) || + return -1; + } + + rdd_config.max_bs = bs; + } + break; + + case 'r': + { + char *tmp = NULL; + long iters = 0; + iters = strtol (arg, &tmp, 10); + if ((iters == LONG_MAX) || + (iters == LONG_MIN) || (tmp && *tmp)) { - fprintf (stderr, "invalid argument for iterations" + fprintf (stderr, "invalid argument for iterations" "(%s)\n", arg); - return -1; - } - - rdd_config.iters = iters; - } - break; - - case 'm': - { - char *tmp = NULL; - long max_ops = 0; - max_ops = strtol (arg, &tmp, 10); - if ((max_ops == LONG_MAX) || - (max_ops == LONG_MIN) || + return -1; + } + + rdd_config.iters = iters; + } + break; + + case 'm': + { + char *tmp = NULL; + long max_ops = 0; + max_ops = strtol (arg, &tmp, 10); + if ((max_ops == LONG_MAX) || + (max_ops == LONG_MIN) || (tmp && *tmp)) { - fprintf (stderr, "invalid argument for max-ops" + fprintf (stderr, "invalid argument for max-ops" "(%s)\n", arg); - return -1; - } - - rdd_config.max_ops_per_seq = max_ops; - } - break; - - case 't': - { - char *tmp = NULL; - long threads = 0; - threads = strtol (arg, &tmp, 10); - if ((threads == LONG_MAX) || - (threads == LONG_MIN) || + return -1; + } + + rdd_config.max_ops_per_seq = max_ops; + } + break; + + case 't': + { + char *tmp = NULL; + long threads = 0; + threads = strtol (arg, &tmp, 10); + if ((threads == LONG_MAX) || + (threads == LONG_MIN) || (tmp && *tmp)) { - fprintf (stderr, "invalid argument for thread count" + fprintf (stderr, "invalid argument for thread count" "(%s)\n", arg); - return -1; - } + return -1; + } - rdd_config.thread_count = threads; - } - break; + rdd_config.thread_count = threads; + } + break; - case ARGP_KEY_NO_ARGS: - break; - case ARGP_KEY_ARG: - break; - case ARGP_KEY_END: - if (_state->argc == 1) { - argp_usage (_state); - } + case ARGP_KEY_NO_ARGS: + break; + case ARGP_KEY_ARG: + break; + case ARGP_KEY_END: + if (_state->argc == 1) { + argp_usage (_state); + } - } + } - return 0; + return 0; +} + +int +string2bytesize (const char *str, unsigned long long *n) +{ + unsigned long long value = 0ULL; + char *tail = NULL; + int old_errno = 0; + const char *s = NULL; + + if (str == NULL || n == NULL) + { + errno = EINVAL; + return -1; + } + + for (s = str; *s != '\0'; s++) + { + if (isspace (*s)) + { + continue; + } + if (*s == '-') + { + return -1; + } + break; + } + + old_errno = errno; + errno = 0; + value = strtoull (str, &tail, 10); + + if (errno == ERANGE || errno == EINVAL) + { + return -1; + } + + if (errno == 0) + { + errno = old_errno; + } + + if (tail[0] != '\0') + { + if (strcasecmp (tail, UNIT_KB_STRING) == 0) + { + value *= UNIT_KB; + } + else if (strcasecmp (tail, UNIT_MB_STRING) == 0) + { + value *= UNIT_MB; + } + else if (strcasecmp (tail, UNIT_GB_STRING) == 0) + { + value *= UNIT_GB; + } + else if (strcasecmp (tail, UNIT_TB_STRING) == 0) + { + value *= UNIT_TB; + } + else if (strcasecmp (tail, UNIT_PB_STRING) == 0) + { + value *= UNIT_PB; + } + + else + { + return -1; + } + } + + *n = value; + + return 0; } static struct argp_option rdd_options[] = { - {"if", 'i', "INPUT_FILE", 0, "input-file"}, - {"of", 'o', "OUTPUT_FILE", 0, "output-file"}, - {"threads", 't', "COUNT", 0, "number of threads to spawn (defaults to 2)"}, - {"min-bs", RDD_MIN_BS_KEY, "MIN_BLOCK_SIZE", 0, - "Minimum block size in bytes (defaults to 1024)"}, - {"max-bs", RDD_MAX_BS_KEY, "MAX_BLOCK_SIZE", 0, - "Maximum block size in bytes (defaults to 4096)"}, - {"iters", 'r', "ITERS", 0, - "Number of read-write sequences (defaults to 1000000)"}, - {"max-ops", 'm', "MAXOPS", 0, - "maximum number of read-writes to be performed in a sequence (defaults to 1)"}, - {0, 0, 0, 0, 0} + {"if", 'i', "INPUT_FILE", 0, "input-file"}, + {"of", 'o', "OUTPUT_FILE", 0, "output-file"}, + {"threads", 't', "COUNT", 0, "number of threads to spawn (defaults to 2)"}, + {"min-bs", RDD_MIN_BS_KEY, "MIN_BLOCK_SIZE", 0, + "Minimum block size in bytes (defaults to 1024)"}, + {"max-bs", RDD_MAX_BS_KEY, "MAX_BLOCK_SIZE", 0, + "Maximum block size in bytes (defaults to 4096)"}, + {"iters", 'r', "ITERS", 0, + "Number of read-write sequences (defaults to 1000000)"}, + {"max-ops", 'm', "MAXOPS", 0, + "maximum number of read-writes to be performed in a sequence (defaults to 1)"}, + {"file-size", 'f', "FILESIZE", 0, + "the size of the file which will be created and upon it I/O will be done" + " (defaults to 100MB"}, + {0, 0, 0, 0, 0} }; static struct argp argp = { @@ -216,277 +312,341 @@ static struct argp argp = { static void rdd_default_config (void) { - rdd_config.thread_count = 2; - rdd_config.iters = 1000000; - rdd_config.max_bs = 4096; - rdd_config.min_bs = 1024; - rdd_config.in_file.fd = rdd_config.out_file.fd = -1; - rdd_config.max_ops_per_seq = 1; - - return; + char *tmp_path = "rdd.in"; + + rdd_config.thread_count = 2; + rdd_config.iters = 1000000; + rdd_config.max_bs = 4096; + rdd_config.min_bs = 1024; + rdd_config.in_file.fd = rdd_config.out_file.fd = -1; + rdd_config.max_ops_per_seq = 1; + strncpy (rdd_config.in_file.path, tmp_path, strlen (tmp_path)); + rdd_config.file_size = 104857600; + + return; } static char rdd_valid_config (void) { - char ret = 1; - int fd = -1; - - fd = open (rdd_config.in_file.path, O_RDONLY); - if (fd == -1) { - ret = 0; - goto out; - } - close (fd); - - if (rdd_config.min_bs > rdd_config.max_bs) { - ret = 0; - goto out; - } - - if (strlen (rdd_config.out_file.path) == 0) { - sprintf (rdd_config.out_file.path, "%s.rddout", + char ret = 1; + int fd = -1; + + fd = open (rdd_config.in_file.path, O_RDONLY); + if (fd == -1 && (errno != ENOENT)) { + fprintf (stderr, "open: (%s)", strerror (errno)); + ret = 0; + goto out; + } + close (fd); + + if (rdd_config.min_bs > rdd_config.max_bs) { + fprintf (stderr, "minimum blocksize %ld is greater than the " + "maximum blocksize %ld", rdd_config.min_bs, + rdd_config.max_bs); + ret = 0; + goto out; + } + + if (strlen (rdd_config.out_file.path) == 0) { + sprintf (rdd_config.out_file.path, "%s.rddout", rdd_config.in_file.path); - } + } out: - return ret; + return ret; } static void * rdd_read_write (void *arg) { - int i = 0, ret = 0; - size_t bs = 0; - off_t offset = 0; - long rand = 0; - long max_ops = 0; - char *buf = NULL; - - buf = calloc (1, rdd_config.max_bs); - if (!buf) { - fprintf (stderr, "calloc failed (%s)\n", strerror (errno)); - ret = -1; - goto out; - } - - for (i = 0; i < rdd_config.iters; i++) - { - pthread_mutex_lock (&rdd_config.lock); - { - int bytes = 0; - rand = random (); - - if (rdd_config.min_bs == rdd_config.max_bs) { - bs = rdd_config.max_bs; - } else { - bs = rdd_config.min_bs + - (rand % - (rdd_config.max_bs - + int i = 0, ret = 0; + size_t bs = 0; + off_t offset = 0; + long rand = 0; + long max_ops = 0; + char *buf = NULL; + + buf = calloc (1, rdd_config.max_bs); + if (!buf) { + fprintf (stderr, "calloc failed (%s)\n", strerror (errno)); + ret = -1; + goto out; + } + + for (i = 0; i < rdd_config.iters; i++) + { + pthread_mutex_lock (&rdd_config.lock); + { + int bytes = 0; + rand = random (); + + if (rdd_config.min_bs == rdd_config.max_bs) { + bs = rdd_config.max_bs; + } else { + bs = rdd_config.min_bs + + (rand % + (rdd_config.max_bs - rdd_config.min_bs)); - } - - offset = rand % rdd_config.in_file.st.st_size; - max_ops = rand % rdd_config.max_ops_per_seq; - if (!max_ops) { - max_ops ++; - } - - ret = lseek (rdd_config.in_file.fd, offset, SEEK_SET); - if (ret != offset) { - fprintf (stderr, "lseek failed (%s)\n", + } + + offset = rand % rdd_config.in_file.st.st_size; + max_ops = rand % rdd_config.max_ops_per_seq; + if (!max_ops) { + max_ops ++; + } + + ret = lseek (rdd_config.in_file.fd, offset, SEEK_SET); + if (ret != offset) { + fprintf (stderr, "lseek failed (%s)\n", strerror (errno)); - ret = -1; - goto unlock; - } + ret = -1; + goto unlock; + } - ret = lseek (rdd_config.out_file.fd, offset, SEEK_SET); - if (ret != offset) { - fprintf (stderr, "lseek failed (%s)\n", + ret = lseek (rdd_config.out_file.fd, offset, SEEK_SET); + if (ret != offset) { + fprintf (stderr, "lseek failed (%s)\n", strerror (errno)); - ret = -1; - goto unlock; - } - - while (max_ops--) - { - bytes = read (rdd_config.in_file.fd, buf, bs); - if (!bytes) { - break; - } - - if (bytes == -1) { - fprintf (stderr, "read failed (%s)\n", + ret = -1; + goto unlock; + } + + while (max_ops--) + { + bytes = read (rdd_config.in_file.fd, buf, bs); + if (!bytes) { + break; + } + + if (bytes == -1) { + fprintf (stderr, "read failed (%s)\n", strerror (errno)); - ret = -1; - goto unlock; - } + ret = -1; + goto unlock; + } - if (write (rdd_config.out_file.fd, buf, bytes) + if (write (rdd_config.out_file.fd, buf, bytes) != bytes) { - fprintf (stderr, "write failed (%s)\n", + fprintf (stderr, "write failed (%s)\n", strerror (errno)); - ret = -1; - goto unlock; - } - } - } - unlock: - pthread_mutex_unlock (&rdd_config.lock); - if (ret == -1) { - goto out; - } - ret = 0; - } + ret = -1; + goto unlock; + } + } + } + unlock: + pthread_mutex_unlock (&rdd_config.lock); + if (ret == -1) { + goto out; + } + ret = 0; + } out: - free (buf); - pthread_barrier_wait (&rdd_config.barrier); + free (buf); + pthread_barrier_wait (&rdd_config.barrier); - return NULL; + return NULL; } +static void +cleanup (void) +{ + close (rdd_config.in_file.fd); + close (rdd_config.out_file.fd); + rdd_config.in_file.fd = rdd_config.out_file.fd = -1; +} + +static int +check_and_create (void) +{ + int ret = -1; + char buf[4096] = {0,}; + struct stat stbuf = {0,}; + int fd[2] = {-1,}; + size_t total_size = -1; + + total_size = rdd_config.file_size; + + ret = stat (rdd_config.in_file.path, &stbuf); + if (ret == -1 && (errno != ENOENT)) + goto out; + + fd[1] = open (rdd_config.in_file.path, O_CREAT | O_WRONLY | O_TRUNC); + if (fd[1] == -1) + goto out; + + fd[0] = open ("/dev/urandom", O_RDONLY); + if (fd[0] == -1) + goto out; + + while (total_size > 0) { + if (total_size >= 4096) { + ret = read (fd[0], buf, 4096); + if (ret == -1) + goto out; + ret = write (fd[1], buf, 4096); + if (ret == -1) + goto out; + total_size = total_size - 4096; + } else { + ret = read (fd[0], buf, total_size); + if (ret == -1) + goto out; + ret = write (fd[1], buf, total_size); + if (ret == -1) + goto out; + total_size = total_size - total_size; + } + + } + + ret = 0; + +out: + if (fd[0] > 0) + close (fd[0]); + if (fd[1] > 0) + close (fd[1]); + return ret; +} static int rdd_spawn_threads (void) { - int i = 0, ret = -1, fd = -1; - char buf[4096]; + int i = 0, ret = -1, fd = -1; + char buf[4096]; - fd = open (rdd_config.in_file.path, O_RDONLY); - if (fd < 0) { - fprintf (stderr, "cannot open %s (%s)\n", + ret = check_and_create (); + if (ret == -1) + goto out; + + fd = open (rdd_config.in_file.path, O_RDONLY); + if (fd < 0) { + fprintf (stderr, "cannot open %s (%s)\n", rdd_config.in_file.path, strerror (errno)); - ret = -1; - goto out; - } - ret = fstat (fd, &rdd_config.in_file.st); - if (ret != 0) { - close (fd); - fprintf (stderr, "cannot stat %s (%s)\n", + ret = -1; + goto out; + } + ret = fstat (fd, &rdd_config.in_file.st); + if (ret != 0) { + close (fd); + fprintf (stderr, "cannot stat %s (%s)\n", rdd_config.in_file.path, strerror (errno)); - ret = -1; - goto out; - } - rdd_config.in_file.fd = fd; + ret = -1; + goto out; + } + rdd_config.in_file.fd = fd; - fd = open (rdd_config.out_file.path, O_WRONLY | O_CREAT, + fd = open (rdd_config.out_file.path, O_WRONLY | O_CREAT | O_TRUNC, S_IRWXU | S_IROTH); - if (fd < 0) { - close (rdd_config.in_file.fd); - rdd_config.in_file.fd = -1; - fprintf (stderr, "cannot open %s (%s)\n", + if (fd < 0) { + close (rdd_config.in_file.fd); + rdd_config.in_file.fd = -1; + fprintf (stderr, "cannot open %s (%s)\n", rdd_config.out_file.path, strerror (errno)); - ret = -1; - goto out; - } - rdd_config.out_file.fd = fd; - - while ((ret = read (rdd_config.in_file.fd, buf, 4096)) > 0) { - if (write (rdd_config.out_file.fd, buf, ret) != ret) { - fprintf (stderr, "write failed (%s)\n", + ret = -1; + goto out; + } + rdd_config.out_file.fd = fd; + + while ((ret = read (rdd_config.in_file.fd, buf, 4096)) > 0) { + if (write (rdd_config.out_file.fd, buf, ret) != ret) { + fprintf (stderr, "write failed (%s)\n", strerror (errno)); - close (rdd_config.in_file.fd); - close (rdd_config.out_file.fd); - rdd_config.in_file.fd = rdd_config.out_file.fd = -1; - ret = -1; - goto out; - } - } - - rdd_config.threads = calloc (rdd_config.thread_count, + cleanup (); + ret = -1; + goto out; + } + } + + rdd_config.threads = calloc (rdd_config.thread_count, sizeof (pthread_t)); - if (rdd_config.threads == NULL) { - fprintf (stderr, "calloc() failed (%s)\n", strerror (errno)); + if (rdd_config.threads == NULL) { + fprintf (stderr, "calloc() failed (%s)\n", strerror (errno)); - ret = -1; - close (rdd_config.in_file.fd); - close (rdd_config.out_file.fd); - rdd_config.in_file.fd = rdd_config.out_file.fd = -1; - goto out; - } + ret = -1; + cleanup (); + goto out; + } - ret = pthread_barrier_init (&rdd_config.barrier, NULL, + ret = pthread_barrier_init (&rdd_config.barrier, NULL, rdd_config.thread_count + 1); - if (ret != 0) { - fprintf (stderr, "pthread_barrier_init() failed (%s)\n", + if (ret != 0) { + fprintf (stderr, "pthread_barrier_init() failed (%s)\n", strerror (ret)); - free (rdd_config.threads); - close (rdd_config.in_file.fd); - close (rdd_config.out_file.fd); - rdd_config.in_file.fd = rdd_config.out_file.fd = -1; - ret = -1; - goto out; - } - - ret = pthread_mutex_init (&rdd_config.lock, NULL); - if (ret != 0) { - fprintf (stderr, "pthread_mutex_init() failed (%s)\n", + free (rdd_config.threads); + cleanup (); + ret = -1; + goto out; + } + + ret = pthread_mutex_init (&rdd_config.lock, NULL); + if (ret != 0) { + fprintf (stderr, "pthread_mutex_init() failed (%s)\n", strerror (ret)); - free (rdd_config.threads); - pthread_barrier_destroy (&rdd_config.barrier); - close (rdd_config.in_file.fd); - close (rdd_config.out_file.fd); - rdd_config.in_file.fd = rdd_config.out_file.fd = -1; - ret = -1; - goto out; - } - - for (i = 0; i < rdd_config.thread_count; i++) - { - ret = pthread_create (&rdd_config.threads[i], NULL, + free (rdd_config.threads); + pthread_barrier_destroy (&rdd_config.barrier); + cleanup (); + ret = -1; + goto out; + } + + for (i = 0; i < rdd_config.thread_count; i++) + { + ret = pthread_create (&rdd_config.threads[i], NULL, rdd_read_write, NULL); - if (ret != 0) { - fprintf (stderr, "pthread_create failed (%s)\n", + if (ret != 0) { + fprintf (stderr, "pthread_create failed (%s)\n", strerror (errno)); - exit (1); - } - } + exit (1); + } + } out: - return ret; + return ret; } - static void rdd_wait_for_completion (void) { - pthread_barrier_wait (&rdd_config.barrier); + pthread_barrier_wait (&rdd_config.barrier); } -int +int main (int argc, char *argv[]) { - int ret = -1; + int ret = -1; - rdd_default_config (); + rdd_default_config (); - ret = argp_parse (&argp, argc, argv, 0, 0, NULL); - if (ret != 0) { - ret = -1; - fprintf (stderr, "%s: argp_parse() failed\n", argv[0]); - goto err; - } + ret = argp_parse (&argp, argc, argv, 0, 0, NULL); + if (ret != 0) { + ret = -1; + fprintf (stderr, "%s: argp_parse() failed\n", argv[0]); + goto err; + } - if (!rdd_valid_config ()) { - ret = -1; - fprintf (stderr, "%s: configuration validation failed\n", + if (!rdd_valid_config ()) { + ret = -1; + fprintf (stderr, "%s: configuration validation failed\n", argv[0]); - goto err; - } + goto err; + } - ret = rdd_spawn_threads (); - if (ret != 0) { - fprintf (stderr, "%s: spawning threads failed\n", argv[0]); - goto err; - } + ret = rdd_spawn_threads (); + if (ret != 0) { + fprintf (stderr, "%s: spawning threads failed\n", argv[0]); + goto err; + } - rdd_wait_for_completion (); + rdd_wait_for_completion (); err: - return ret; -} + return ret; +} diff --git a/extras/clear_xattrs.sh b/extras/clear_xattrs.sh new file mode 100755 index 000000000..dd04731e8 --- /dev/null +++ b/extras/clear_xattrs.sh @@ -0,0 +1,53 @@ +#!/bin/bash + +# Clear the trusted.gfid xattr in the brick tree + +# This script must be run only on a stopped brick/volume +# Stop the volume to make sure no rebalance/replace-brick +# operations are on-going + +# Not much error checking +remove_xattrs () +{ + find "$1" -exec setfattr -h -x "trusted.gfid" '{}' \; > /dev/null 2>&1; + find "$1" -exec setfattr -h -x "trusted.glusterfs.volume-id" '{}' \; > /dev/null 2>&1; +} + +main () +{ + if [ -z "$1" ]; then + echo "Usage: $0 <brick_path(s)>"; + exit 1; + fi + + export PATH; + which getfattr > /dev/null 2>&1; + if [ $? -ne 0 ]; then + echo "attr package missing"; + exit 2; + fi + + which setfattr > /dev/null 2>&1; + if [ $? -ne 0 ]; then + echo "attr package missing"; + exit 2; + fi + + for brick in "$@"; + do + stat "$brick" > /dev/null 2>&1; + if [ $? -ne 0 ]; then + echo "brick: $brick does not exist"; + exit 3; + fi + if [ ! -d "$brick" ]; then + echo "$brick: not a directory"; + exit 4; + fi + echo "xattr clean-up in progress: $brick"; + remove_xattrs "$brick"; + echo "$brick ready to be used as a glusterfs brick"; + done; +} + +main "$@";
\ No newline at end of file diff --git a/extras/contri-add.sh b/extras/contri-add.sh new file mode 100755 index 000000000..7db5edd5d --- /dev/null +++ b/extras/contri-add.sh @@ -0,0 +1,73 @@ +#!/bin/bash + +# This script adds contributions of files/directories in backend to volume +# size. +# It can also be used to debug by passing dir as first argument, in which case +# it will just add contributions from immediate children of a directory and +# displays only if added contributions from immediate children is different +# from size stored in directory. +# For Eg., find <backend-directory> -type d -exec ./contri-add.sh dir \{} \; +# will list all the directories which have descrepancies in their +# size/contributions. + +usage () +{ + echo >&2 "usage: $0 <file|dir> <list-of-backend-directories>" +} + +add_contributions () +{ + local var=0 + local count=0 + + SIZE=`getfattr -h -e hex -n trusted.glusterfs.quota.size $2 2>&1 | sed -e '/^#/d' | sed -e '/^getfattr/d' | sed -e '/^$/d' | cut -d'=' -f 2` + CONTRI=`getfattr -h -e hex -d -m trusted.glusterfs.quota.*.contri $2 2>&1 | sed -e '/^#/d' | sed -e '/^getfattr/d' | sed -e '/^$/d' | cut -d'=' -f 2` + + if [ $1 == "file" ]; then + PATHS=`find $2 ! -type d | sed -e "\|^$2$|d" | sed -e '/^[ \t]*$/d'` + else + PATHS=`find $2 -maxdepth 1 | sed -e "\|^$2$|d" | sed -e '/^[ \t]*$/d'` + fi + + if [ -z "$PATHS" ]; then + return 0 + fi + + CONTRIBUTIONS=`echo $PATHS | xargs getfattr -h -e hex -d -m trusted.glusterfs.quota.*.contri 2>&1 | sed -e '/^#/d' | sed -e '/^getfattr/d' | sed -e '/^$/d' | cut -d'=' -f 2 | sed -e 's/^[ \t]*\([^ \t]*\)/\1/g'` + + if [ -n "$CONTRIBUTIONS" ]; then + for i in $CONTRIBUTIONS; do + count=$(($count + 1)) + var=$(($var + $i)) + done + fi + + if [ $1 == "file" ] || [ $var -ne $(($SIZE)) ] || [ $(($SIZE)) -ne $(($CONTRI)) ]; then + if [ $1 == "dir" ]; then + TMP_PATH=`echo $2 | sed -e "s/\/home\/export\/[0-9]*/\/mnt\/raghu/g"` + stat $TMP_PATH > /dev/null + fi + + echo "file count $count" + echo "added contribution of $2=$var" + echo "size stored in xattrs on $2=$(($SIZE))" + echo "contribution of $2 to its parent directory=$(($CONTRI))" + echo "==============================================================" + fi +} + + +main () +{ + [ $# -lt 1 ] && usage + + TYPE=$1 + + shift 1 + + for i in $@; do + add_contributions $TYPE $i + done +} + +main $@
\ No newline at end of file diff --git a/extras/disk_usage_sync.sh b/extras/disk_usage_sync.sh new file mode 100755 index 000000000..85ee158f8 --- /dev/null +++ b/extras/disk_usage_sync.sh @@ -0,0 +1,86 @@ +#!/bin/sh + +# This script can be used to sync disk usage before activating quotas on GlusterFS. +# There are two scenarios where quotas are used and this script has to be used accordingly - +# +# 1. Server side +# The script needs to be run with backend export directory as the argument. This script updates +# the current disk usage of the backend directory in a way that is intelligible to the GlusterFS quota +# translator. Make sure you run this script before starting glusterfsd (GlusterFS server process): +# * for the first time +# * after any server outage (reboot, etc.) + +# 2. Client side +# The script needs to be run with the client mount point as the argument. It updates the current disk +# of the GlusterFS volume is a way that is intelligible to the GlusterFS quota translator. Make sure +# you run this script after a fresh mount of the GlusterFS volume on the client: +# * For the first time +# * After any client outage (reboot, remount, etc.) + +# Please note that this script is dependent on the 'attr' package, more specifically 'setfattr' to set +# extended attributes of files. +# GlusterFS +# +# (c) 2010 Gluster Inc <http://www.gluster.com/> + +PROGRAM_NAME="disk_usage_sync.sh" + +#check if setfattr is available +check_for_attr () +{ + `command -v setfattr >/dev/null` + if [ "${?}" -gt 0 ]; then + echo >&2 "This script requires the 'attr' package to run. Either it has not been installed or is not present currently in the system path." + exit 1 + fi + +} + +usage () { + echo >&2 "$PROGRAM_NAME - Command used to sync disk usage information before activating quotas on GlusterFS. + +usage: $PROGRAM_NAME <target-directory>" + +exit 1 +} + +TARGET_DIR=$1 +EXT_ATTR_NAME="trusted.glusterfs-quota-du" + +#output of du in number of bytes +get_disk_usage () +{ + if [ ! -d $TARGET_DIR ]; then + echo >&2 "Error: $TARGET_DIR does not exist." + exit 1 + fi + + DISK_USAGE=`du -bc $TARGET_DIR | grep 'total' | cut -f1` + if [ "${?}" -gt 0 ]; then + exit 1 + fi + +} + +#set the extended attribute of the root directory with du size in bytes +set_disk_usage () +{ + ` setfattr -n $EXT_ATTR_NAME -v $DISK_USAGE $TARGET_DIR` + if [ "${?}" -gt 0 ]; then + exit 1 + fi +} + +main () +{ + [ $# -lt 1 ] && usage + + check_for_attr + + get_disk_usage + set_disk_usage + + printf "Disk Usage information has been sync'd successfully.\n" +} + +main "$@" diff --git a/extras/file_size_contri.sh b/extras/file_size_contri.sh new file mode 100755 index 000000000..4f52a9a89 --- /dev/null +++ b/extras/file_size_contri.sh @@ -0,0 +1,17 @@ +#!/bin/bash + +# This script checks whether the contribution and disk-usage of a file is same. + +CONTRIBUTION_HEX=`getfattr -h -e hex -d -m trusted.glusterfs.quota.*.contri $1 2>&1 | sed -e '/^#/d' | sed -e '/^getfattr/d' | sed -e '/^$/d' | cut -d'=' -f 2` + +BLOCKS=`stat -c %b $1` +SIZE=$(($BLOCKS * 512)) + +CONTRIBUTION=`printf "%d" $CONTRIBUTION_HEX` + +if [ $CONTRIBUTION -ne $SIZE ]; then + printf "contribution of %s:%d\n" $1 $CONTRIBUTION + echo "size of $1: $SIZE" + echo "===================================================" +fi + diff --git a/extras/generate-xdr-files.sh b/extras/generate-xdr-files.sh new file mode 100755 index 000000000..bc02f77c9 --- /dev/null +++ b/extras/generate-xdr-files.sh @@ -0,0 +1,98 @@ +#!/bin/sh + +_init () +{ + xfile="$1"; + # TODO: check the validity of .x file + + cfile="${1%.x}.c"; + hfile="${1%.x}.h"; + + tmp_cfile="$1.c"; + + tmp1_hfile="$1.h.tmp"; + tmp1_cfile="$1.c.tmp"; + +} + +append_licence_header () +{ + src_file=$1; + dst_file=$2; + + cat >$dst_file <<EOF +/* + Copyright (c) 2007-2012 Red Hat, Inc. <http://www.redhat.com> + This file is part of GlusterFS. + + This file is licensed to you under your choice of the GNU Lesser + General Public License, version 3 or any later version (LGPLv3 or + later), or the GNU General Public License, version 2 (GPLv2), in all + cases as published by the Free Software Foundation. +*/ + +#include "xdr-common.h" +#include "compat.h" + +#if defined(__GNUC__) +#if __GNUC__ >= 4 +#pragma GCC diagnostic ignored "-Wunused-but-set-variable" +#endif +#endif + +EOF + + cat $src_file >> $dst_file; + +} + +main () +{ + if [ $# -ne 1 ]; then + echo "wrong number of arguments given" + echo " $0 <XDR-definition-file>.x" + exit 1; + fi + + + echo -n "writing the XDR routine file ($tmp_cfile) ... "; + rm -f $tmp_cfile; + rpcgen -c -o $tmp_cfile $xfile; + + # get rid of warnings in xdr .c file due to "unused variable 'buf'" + sed -i -e 's:buf;$:buf;\ + buf = NULL;:' $tmp_cfile; + + sed -i '/int i;/d' $tmp_cfile; + + echo "OK"; + + # no need for a temporary file here as there are no changes from glusterfs + echo -n "writing the XDR header file ($hfile) ... "; + rm -f $hfile; + rpcgen -h -o $hfile $xfile; + + # the '#ifdef' part of file should be fixed + sed -i -e 's/-/_/g' $hfile; + + echo "OK"; + + echo -n "writing licence header to the generated files ... "; + # Write header to temp file and append generated file + append_licence_header $hfile $tmp1_hfile; + append_licence_header $tmp_cfile $tmp1_cfile; + echo "OK" + + # now move the destination file to actual original file + echo -n "updating existing files ... "; + mv $tmp1_hfile $hfile; + mv $tmp1_cfile $cfile; + + # remove unwanted temporary files (if any) + rm -f $tmp_cfile $tmp1_cfile $tmp1_hfile + + echo "OK" + +} + +_init "$@" && main "$@"; diff --git a/extras/geo-rep/Makefile.am b/extras/geo-rep/Makefile.am new file mode 100644 index 000000000..fc5f56d54 --- /dev/null +++ b/extras/geo-rep/Makefile.am @@ -0,0 +1,2 @@ +EXTRA_DIST = gsync-sync-gfid.c gsync-upgrade.sh generate-gfid-file.sh \ + get-gfid.sh slave-upgrade.sh diff --git a/extras/geo-rep/generate-gfid-file.sh b/extras/geo-rep/generate-gfid-file.sh new file mode 100644 index 000000000..c6739fbf1 --- /dev/null +++ b/extras/geo-rep/generate-gfid-file.sh @@ -0,0 +1,53 @@ +#!/bin/bash +#Usage: generate-gfid-file.sh <master-volfile-server:master-volume> <path-to-get-gfid.sh> <output-file> + +function get_gfids() +{ + GET_GFID_CMD=$1 + OUTPUT_FILE=$2 + find . -exec $GET_GFID_CMD {} \; > $OUTPUT_FILE +} + +function mount_client() +{ + local T; # temporary mount + local i; # inode number + + VOLFILE_SERVER=$1; + VOLUME=$2; + GFID_CMD=$3; + OUTPUT=$4; + + T=$(mktemp -d); + + glusterfs -s $VOLFILE_SERVER --volfile-id $VOLUME $T; + + i=$(stat -c '%i' $T); + + [ "x$i" = "x1" ] || fatal "could not mount volume $MASTER on $T"; + + cd $T; + + get_gfids $GFID_CMD $OUTPUT + + cd -; + + umount $T || fatal "could not umount $MASTER from $T"; + + rmdir $T || warn "rmdir of $T failed"; +} + + +function main() +{ + SLAVE=$1 + GET_GFID_CMD=$2 + OUTPUT=$3 + + VOLFILE_SERVER=`echo $SLAVE | sed -e 's/\(.*\):.*/\1/'` + VOLUME_NAME=`echo $SLAVE | sed -e 's/.*:\(.*\)/\1/'` + + mount_client $VOLFILE_SERVER $VOLUME_NAME $GET_GFID_CMD $OUTPUT +} + +main "$@"; diff --git a/extras/geo-rep/get-gfid.sh b/extras/geo-rep/get-gfid.sh new file mode 100755 index 000000000..a4d609b0b --- /dev/null +++ b/extras/geo-rep/get-gfid.sh @@ -0,0 +1,7 @@ +#!/bin/bash + +ATTR_STR=`getfattr -h $1 -n glusterfs.gfid.string` +GLFS_PATH=`echo $ATTR_STR | sed -e 's/# file: \(.*\) glusterfs.gfid.string*/\1/g'` +GFID=`echo $ATTR_STR | sed -e 's/.*glusterfs.gfid.string="\(.*\)"/\1/g'` + +echo "$GFID $GLFS_PATH" diff --git a/extras/geo-rep/gsync-sync-gfid.c b/extras/geo-rep/gsync-sync-gfid.c new file mode 100644 index 000000000..601f4720e --- /dev/null +++ b/extras/geo-rep/gsync-sync-gfid.c @@ -0,0 +1,106 @@ + +#include <stdio.h> +#include <errno.h> +#include <string.h> +#include <limits.h> +#include <sys/types.h> +#include <attr/xattr.h> +#include <libgen.h> +#include <ctype.h> +#include <stdlib.h> + +#ifndef UUID_CANONICAL_FORM_LEN +#define UUID_CANONICAL_FORM_LEN 36 +#endif + +#ifndef GF_FUSE_AUX_GFID_HEAL +#define GF_FUSE_AUX_GFID_HEAL "glusterfs.gfid.heal" +#endif + +#define GLFS_LINE_MAX (PATH_MAX + (2 * UUID_CANONICAL_FORM_LEN)) + +int +main (int argc, char *argv[]) +{ + char *file = NULL; + char *tmp = NULL; + char *tmp1 = NULL; + char *parent_dir = NULL; + char *gfid = NULL; + char *bname = NULL; + int ret = -1; + int len = 0; + FILE *fp = NULL; + char line[GLFS_LINE_MAX] = {0,}; + char *path = NULL; + void *blob = NULL; + void *tmp_blob = NULL; + + if (argc != 2) { + /* each line in the file has the following format + * uuid-in-canonical-form path-relative-to-gluster-mount. + * Both uuid and relative path are from master mount. + */ + fprintf (stderr, "usage: %s <file-of-paths-to-be-synced>\n", + argv[0]); + goto out; + } + + file = argv[1]; + + fp = fopen (file, "r"); + if (fp == NULL) { + fprintf (stderr, "cannot open %s for reading (%s)\n", + file, strerror (errno)); + goto out; + } + + while (fgets (line, GLFS_LINE_MAX, fp) != NULL) { + tmp = line; + path = gfid = line; + + path += UUID_CANONICAL_FORM_LEN + 1; + + while(isspace (*path)) + path++; + + if ((strlen (line) < GLFS_LINE_MAX) && + (line[strlen (line) - 1] == '\n')) + line[strlen (line) - 1] = '\0'; + + line[UUID_CANONICAL_FORM_LEN] = '\0'; + + tmp = strdup (path); + tmp1 = strdup (path); + parent_dir = dirname (tmp); + bname = basename (tmp1); + + /* gfid + '\0' + bname + '\0' */ + len = UUID_CANONICAL_FORM_LEN + 1 + strlen (bname) + 1; + + blob = calloc (1, len); + + memcpy (blob, gfid, UUID_CANONICAL_FORM_LEN); + + tmp_blob = blob + UUID_CANONICAL_FORM_LEN + 1; + + memcpy (tmp_blob, bname, strlen (bname)); + + ret = setxattr (parent_dir, GF_FUSE_AUX_GFID_HEAL, blob, len, + 0); + if (ret < 0) { + fprintf (stderr, "setxattr on %s/%s failed (%s)\n", + parent_dir, bname, strerror (errno)); + } + memset (line, 0, GLFS_LINE_MAX); + + free (blob); + free (tmp); free (tmp1); + blob = NULL; + } + + ret = 0; +out: + return ret; +} + diff --git a/extras/geo-rep/gsync-upgrade.sh b/extras/geo-rep/gsync-upgrade.sh new file mode 100644 index 000000000..b17948736 --- /dev/null +++ b/extras/geo-rep/gsync-upgrade.sh @@ -0,0 +1,123 @@ +#!/bin/bash +#usage: gsync-upgrade.sh <slave-volfile-server:slave-volume> <gfid-file> +# <path-to-gsync-sync-gfid> <ssh-identity-file> +#<slave-volfile-server>: a machine on which gluster cli can fetch slave volume info. +# slave-volfile-server defaults to localhost. +# +#<gfid-file>: a file containing paths and their associated gfids +# on master. The paths are relative to master mount point +# (not absolute). An example extract of <gfid-file> can be, +# +# <extract> +# 22114455-57c5-46e9-a783-c40f83a72b09 /dir +# 25772386-3eb8-4550-a802-c3fdc938ca80 /dir/file +# </extract> +# +#<ssh-identity-file>: file from which the identity (private key) for public key authentication is read. + +SLAVE_MOUNT='/tmp/glfs_slave' + +function SSH() +{ + HOST=$1 + SSHKEY=$2 + + shift 2 + + ssh -qi $SSHKEY \ + -oPasswordAuthentication=no \ + -oStrictHostKeyChecking=no \ + "$HOST" "$@"; +} + +function get_bricks() +{ + SSHKEY=$3 + + SSH $1 $SSHKEY "gluster volume info $2" | grep -E 'Brick[0-9]+' | sed -e 's/[^:]*:\(.*\)/\1/g' +} + +function cleanup_brick() +{ + HOST=$1 + BRICK=$2 + SSHKEY=$3 + + # TODO: write a C program to receive a list of files and does cleanup on + # them instead of spawning a new setfattr process for each file if + # performance is bad. + SSH -i $SSHKEY $HOST "rm -rf $BRICK/.glusterfs/* && find $BRICK -exec setfattr -x trusted.gfid {} \;" +} + +function cleanup_slave() +{ + SSHKEY=$2 + + VOLFILE_SERVER=`echo $1 | sed -e 's/\(.*\):.*/\1/'` + VOLUME_NAME=`echo $1 | sed -e 's/.*:\(.*\)/\1/'` + + BRICKS=`get_bricks $VOLFILE_SERVER $VOLUME_NAME $SSHKEY` + + for i in $BRICKS; do + HOST=`echo $i | sed -e 's/\(.*\):.*/\1/'` + BRICK=`echo $i | sed -e 's/.*:\(.*\)/\1/'` + cleanup_brick $HOST $BRICK $SSHKEY + done + + SSH -i $SSHKEY $VOLFILE_SERVER "gluster --mode=script volume stop $VOLUME_NAME; gluster volume start $VOLUME_NAME"; + +} + +function mount_client() +{ + local T; # temporary mount + local i; # inode number + GFID_FILE=$3 + SYNC_CMD=$4 + + T=$(mktemp -d); + + glusterfs --aux-gfid-mount -s $1 --volfile-id $2 $T; + + i=$(stat -c '%i' $T); + + [ "x$i" = "x1" ] || fatal "could not mount volume $MASTER on $T"; + + cd $T; + + $SYNC_CMD $GFID_FILE + + cd -; + + umount -l $T || fatal "could not umount $MASTER from $T"; + + rmdir $T || warn "rmdir of $T failed"; +} + +function sync_gfids() +{ + SLAVE=$1 + GFID_FILE=$2 + + SLAVE_VOLFILE_SERVER=`echo $SLAVE | sed -e 's/\(.*\):.*/\1/'` + SLAVE_VOLUME_NAME=`echo $SLAVE | sed -e 's/.*:\(.*\)/\1/'` + + if [ "x$SLAVE_VOLFILE_SERVER" = "x" ]; then + SLAVE_VOLFILE_SERVER="localhost" + fi + + mount_client $SLAVE_VOLFILE_SERVER $SLAVE_VOLUME_NAME $GFID_FILE $3 +} + +function upgrade() +{ + SLAVE=$1 + GFID_FILE=$2 + SYNC_CMD=$3 + SSHKEY=$4 + + cleanup_slave $SLAVE $SSHKEY + sync_gfids $SLAVE $GFID_FILE $SYNC_CMD +} + +upgrade "$@" diff --git a/extras/geo-rep/slave-upgrade.sh b/extras/geo-rep/slave-upgrade.sh new file mode 100644 index 000000000..6198f408a --- /dev/null +++ b/extras/geo-rep/slave-upgrade.sh @@ -0,0 +1,102 @@ +#!/bin/bash +#usage: slave-upgrade.sh <volfile-server:volname> <gfid-file> +# <path-to-gsync-sync-gfid> +#<slave-volfile-server>: a machine on which gluster cli can fetch slave volume info. +# slave-volfile-server defaults to localhost. +# +#<gfid-file>: a file containing paths and their associated gfids +# on master. The paths are relative to master mount point +# (not absolute). An example extract of <gfid-file> can be, +# +# <extract> +# 22114455-57c5-46e9-a783-c40f83a72b09 /dir +# 25772386-3eb8-4550-a802-c3fdc938ca80 /dir/file +# </extract> + +function get_bricks() +{ + gluster volume info $1 | grep -E 'Brick[0-9]+' | sed -e 's/[^:]*:\(.*\)/\1/g' +} + +function cleanup_brick() +{ + HOST=$1 + BRICK=$2 + + # TODO: write a C program to receive a list of files and does cleanup on + # them instead of spawning a new setfattr process for each file if + # performance is bad. + ssh $HOST "rm -rf $BRICK/.glusterfs/* && find $BRICK -exec setfattr -x trusted.gfid {} \; 2>/dev/null" +} + +function cleanup_slave() +{ + VOLUME_NAME=`echo $1 | sed -e 's/.*:\(.*\)/\1/'` + + BRICKS=`get_bricks $VOLUME_NAME` + + for i in $BRICKS; do + HOST=`echo $i | sed -e 's/\(.*\):.*/\1/'` + BRICK=`echo $i | sed -e 's/.*:\(.*\)/\1/'` + cleanup_brick $HOST $BRICK + done + + # Now restart the volume + gluster --mode=script volume stop $VOLUME_NAME; + gluster volume start $VOLUME_NAME; +} + +function mount_client() +{ + local T; # temporary mount + local i; # inode number + + VOLUME_NAME=$2; + GFID_FILE=$3 + SYNC_CMD=$4 + + T=$(mktemp -d); + + glusterfs --aux-gfid-mount -s $1 --volfile-id $VOLUME_NAME $T; + + i=$(stat -c '%i' $T); + + cd $T; + + $SYNC_CMD $GFID_FILE + + cd -; + + umount $T || fatal "could not umount $MASTER from $T"; + + rmdir $T || warn "rmdir of $T failed"; +} + +function sync_gfids() +{ + SLAVE=$1 + GFID_FILE=$2 + SYNC_CMD=$3 + + SLAVE_VOLFILE_SERVER=`echo $SLAVE | sed -e 's/\(.*\):.*/\1/'` + SLAVE_VOLUME_NAME=`echo $SLAVE | sed -e 's/.*:\(.*\)/\1/'` + + if [ "x$SLAVE_VOLFILE_SERVER" = "x" ]; then + SLAVE_VOLFILE_SERVER="localhost" + fi + + mount_client $SLAVE_VOLFILE_SERVER $SLAVE_VOLUME_NAME $GFID_FILE $SYNC_CMD +} + +function upgrade() +{ + SLAVE=$1 + GFID_FILE=$2 + SYNC_CMD=$3 + + cleanup_slave $SLAVE + + sync_gfids $SLAVE $GFID_FILE $SYNC_CMD +} + +upgrade "$@" diff --git a/extras/gluster-rsyslog-5.8.conf b/extras/gluster-rsyslog-5.8.conf new file mode 100644 index 000000000..2519999bc --- /dev/null +++ b/extras/gluster-rsyslog-5.8.conf @@ -0,0 +1,51 @@ +##### gluster.conf ##### + +# +## If you want to log every message to the log file instead of +## intelligently suppressing repeated messages, set off to +## RepeatedMsgReduction. This change requires rsyslog restart +## (eg. run 'service rsyslog restart') +# +#$RepeatedMsgReduction off +$RepeatedMsgReduction on + +# +## The mmcount module provides the capability to count log messages by +## severity or json property of given app-name. The count value is added +## into the log message as json property named '$msgid' +# +$ModLoad mmcount +$mmcountKey gf_code # start counting value of gf_code + +$template Glusterfsd_dynLogFile,"/var/log/glusterfs/bricks/%app-name%.log" +$template Gluster_dynLogFile,"/var/log/glusterfs/%app-name%.log" + +$template GLFS_Template,"%msgid%/%syslogfacility-text:::uppercase%/%syslogseverity-text:::uppercase% [%TIMESTAMP:::date-rfc3339%] %msg:::sp-if-no-1st-sp%%msg:::drop-last-lf%\n" + +# +## Pass logs to mmcount if app-name is 'gluster' +# +if $app-name contains 'gluster' then :mmcount: + +if $app-name contains 'glusterfsd' then ?Glusterfsd_dynLogFile;GLFS_Template +if $app-name contains 'gluster' and not ( $app-name contains 'glusterfsd' ) then ?Gluster_dynLogFile;GLFS_Template + +# +## Sample configuration to send a email alert for every 50th mmcount +# +#$ModLoad ommail +#$ActionMailSMTPServer smtp.example.com +#$ActionMailFrom rsyslog@example.com +#$ActionMailTo glusteradmin@example.com +#$template mailSubject,"50th message of gf_code=9999 on %hostname%" +#$template mailBody,"RSYSLOG Alert\r\nmsg='%msg%'" +#$ActionMailSubject mailSubject +#$ActionExecOnlyOnceEveryInterval 30 +#if $app-name == 'glusterfsd' and $msgid != 0 and $msgid % 50 == 0 \ +#then :ommail:;RSYSLOG_SyslogProtocol23Format +# + +# +## discard logs where app-name is 'gluster' as we processed already +# +if $app-name contains 'gluster' then ~ diff --git a/extras/gluster-rsyslog-7.2.conf b/extras/gluster-rsyslog-7.2.conf new file mode 100644 index 000000000..8b2841543 --- /dev/null +++ b/extras/gluster-rsyslog-7.2.conf @@ -0,0 +1,76 @@ +##### gluster.conf ##### +# +## If you want to log every message to the log file instead of +## intelligently suppressing repeated messages, set off to +## RepeatedMsgReduction. This change requires rsyslog restart +## (eg. run 'service rsyslog restart') +# +#$RepeatedMsgReduction off +$RepeatedMsgReduction on + +$ModLoad mmjsonparse +*.* :mmjsonparse: + +# +## The mmcount module provides the capability to count log messages by +## severity or json property of given app-name. The count value is added +## into the log message as json property named 'mmcount' +## +## More info at http://www.rsyslog.com/doc/mmcount.html +# +#module(load="mmcount") +#action(type="mmcount" appname="glusterd" key="!gf_code") # count each value of gf_code of appname glusterd +#action(type="mmcount" appname="glusterfsd" key="!gf_code") # count each value of gf_code of appname glusterfsd +#action(type="mmcount" appname="glusterfs" key="!gf_code") # count each value of gf_code of appname glusterfs + +template (name="Glusterfsd_dynLogFile" type="string" string="/var/log/glusterfs/bricks/%app-name%.log") +template (name="Gluster_dynLogFile" type="string" string="/var/log/glusterfs/%app-name%.log") + +template(name="GLFS_template" type="list") { + property(name="$!mmcount") + constant(value="/") + property(name="syslogfacility-text" caseConversion="upper") + constant(value="/") + property(name="syslogseverity-text" caseConversion="upper") + constant(value=" ") + constant(value="[") + property(name="timereported" dateFormat="rfc3339") + constant(value="] ") + constant(value="[") + property(name="$!gf_code") + constant(value="] ") + constant(value="[") + property(name="$!gf_message") + constant(value="] ") + property(name="$!msg") + constant(value="\n") +} + +if $app-name contains 'glusterfsd' then { + action(type="omfile" + DynaFile="Glusterfsd_dynLogFile" + Template="GLFS_template") + stop +} + +if $app-name contains 'gluster' then { + action(type="omfile" + DynaFile="Gluster_dynLogFile" + Template="GLFS_template") + stop +} + +# +## send email for every 50th mmcount +#$ModLoad ommail +#if $app-name == 'glusterfsd' and $!mmcount <> 0 and $!mmcount % 50 == 0 then { +# $ActionMailSMTPServer smtp.example.com +# $ActionMailFrom rsyslog@example.com +# $ActionMailTo glusteradmin@example.com +# $template mailSubject,"50th message of gf_code=9999 on %hostname%" +# $template mailBody,"RSYSLOG Alert\r\nmsg='%msg%'" +# $ActionMailSubject mailSubject +# $ActionExecOnlyOnceEveryInterval 30 +# :ommail:;RSYSLOG_SyslogProtocol23Format +#} +# diff --git a/extras/glusterd-sysconfig b/extras/glusterd-sysconfig new file mode 100644 index 000000000..8237c5711 --- /dev/null +++ b/extras/glusterd-sysconfig @@ -0,0 +1,6 @@ +## Set custom log file and log level (bellow are defaults) +# LOG_FILE='/var/log/glusterfs/glusterd.log' +# LOG_LEVEL='INFO' + +## Set custom options for glusterd +# GLUSTERD_OPTIONS='' diff --git a/extras/glusterd.vol b/extras/glusterd.vol new file mode 100644 index 000000000..9bac52ab7 --- /dev/null +++ b/extras/glusterd.vol @@ -0,0 +1,9 @@ +volume management + type mgmt/glusterd + option working-directory /var/lib/glusterd + option transport-type socket,rdma + option transport.socket.keepalive-time 10 + option transport.socket.keepalive-interval 2 + option transport.socket.read-fail-log off +# option base-port 49152 +end-volume diff --git a/extras/glusterfs-georep-logrotate b/extras/glusterfs-georep-logrotate new file mode 100644 index 000000000..6a69ab1e3 --- /dev/null +++ b/extras/glusterfs-georep-logrotate @@ -0,0 +1,18 @@ + +rotate 52 +missingok + +compress +delaycompress +notifempty + +/var/log/glusterfs/geo-replication/*/*.log { +} + + +/var/log/glusterfs/geo-replication-slaves/*.log { +} + + +/var/log/glusterfs/geo-replication-slaves/*/*.log { +} diff --git a/extras/glusterfs-logrotate b/extras/glusterfs-logrotate new file mode 100644 index 000000000..373ec2e0b --- /dev/null +++ b/extras/glusterfs-logrotate @@ -0,0 +1,27 @@ +# perform the log rotate every week +weekly +# keep the backup of 52 weeks +rotate 52 +missingok + +# compress the logs, but from the .2 onwards +compress +delaycompress +notifempty + +# Rotate client logs +/var/log/glusterfs/*.log { + sharedscripts + postrotate + /usr/bin/killall -HUP glusterfs > /dev/null 2>&1 || true + /usr/bin/killall -HUP glusterd > /dev/null 2>&1 || true + endscript +} + +# Rotate server logs +/var/log/glusterfs/bricks/*.log { + sharedscripts + postrotate + /usr/bin/killall -HUP glusterfsd > /dev/null 2>&1 || true + endscript +} diff --git a/extras/glusterfs-mode.el b/extras/glusterfs-mode.el index e65fbf460..d4f6dc568 100644 --- a/extras/glusterfs-mode.el +++ b/extras/glusterfs-mode.el @@ -1,4 +1,4 @@ -;;; Copyright (C) 2007, 2008 Z RESEARCH Inc. <http://www.zresearch.com>
+;;; Copyright (C) 2007-2011 Gluster Inc. <http://www.gluster.com>
;;;
;;; This program is free software; you can redistribute it and/or modify
;;; it under the terms of the GNU General Public License as published by
diff --git a/extras/glusterfs.vim b/extras/glusterfs.vim index 42fa6b573..899cc6551 100644 --- a/extras/glusterfs.vim +++ b/extras/glusterfs.vim @@ -1,20 +1,11 @@ " glusterfs.vim: GNU Vim Syntax file for GlusterFS .vol specification -" Copyright (C) 2007-2009 Z RESEARCH, Inc. <http://www.zresearch.com> -" This file is part of GlusterFS. +" Copyright (c) 2007777777Red Hat, Inc. <http://www.redhat.com> +" This file is part of GlusterFS. " -" GlusterFS is free software; you can redistribute it and/or modify -" it under the terms of the GNU General Public License as published -" by the Free Software Foundation; either version 3 of the License, -" or (at your option) any later version. -" -" GlusterFS is distributed in the hope that it will be useful, but -" WITHOUT ANY WARRANTY; without even the implied warranty of -" MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU -" General Public License for more details. -" -" You should have received a copy of the GNU General Public License -" along with this program. If not, see -" <http://www.gnu.org/licenses/>. +" This file is licensed to you under your choice of the GNU Lesser +" General Public License, version 3 or any later version (LGPLv3 or +" later), or the GNU General Public License, version 2 (GPLv2), in all +" cases as published by the Free Software Foundation. " " Last Modified: Wed Aug 1 00:47:10 IST 2007 " Version: 0.8 diff --git a/extras/gnfs-loganalyse.py b/extras/gnfs-loganalyse.py new file mode 100644 index 000000000..71e79b6be --- /dev/null +++ b/extras/gnfs-loganalyse.py @@ -0,0 +1,260 @@ +#!/bin/python +""" + Copyright (c) 2006-2012 Red Hat, Inc. <http://www.redhat.com> + This file is part of GlusterFS. + + This file is licensed to you under your choice of the GNU Lesser + General Public License, version 3 or any later version (LGPLv3 or + later), or the GNU General Public License, version 2 (GPLv2), in all + cases as published by the Free Software Foundation. + +""" + +import os +import string +import sys + + +class NFSRequest: + def requestIsEntryOp (self): + op = self.op + if op == "CREATE" or op == "LOOKUP" or op == "REMOVE" or op == "LINK" or op == "RENAME" or op == "MKDIR" or op == "RMDIR" or op == "SYMLINK" or op == "MKNOD": + return 1 + else: + return 0 + + def __init__ (self, logline, linecount): + self.calllinecount = 0 + self.xid = "" + self.op = "" + self.opdata = "" + self.replydata = "" + self.replylinecount = 0 + self.timestamp = "" + self.entryname = "" + self.gfid = "" + self.replygfid = "" + + tokens = logline.strip ().split (" ") + self.timestamp = tokens[0] + " " + tokens[1] + if "XID:" not in tokens: + return None + + if "args:" not in tokens: + return None + + self.calllinecount = linecount + + xididx = tokens.index ("XID:") + self.xid = tokens [xididx+1].strip(",") + + opidx = tokens.index ("args:") + self.op = tokens [opidx-1].strip (":") + self.opdata = " ".join(tokens [opidx+1:]) + if self.requestIsEntryOp (): + nameidx = tokens.index ("name:") + self.entryname = tokens[nameidx + 1].strip (",") + gfididx = tokens.index ("gfid") + self.gfid = tokens[gfididx +1].strip(",") + + + def getXID (self): + return self.xid + + def setReply (self, logline, linecount): + tokens = logline.strip ().split (" ") + timestamp = tokens[0] + " " + tokens[1] + statidx = tokens.index ("NFS:") + self.replydata = " TimeStamp: " + timestamp + " " + " ".join (tokens [statidx+1:]) + self.replylinecount = linecount + if "gfid" in tokens: + gfididx = tokens.index ("gfid") + self.replygfid = tokens [gfididx + 1].strip(",") + + def dump (self): + print "ReqLine: " + str(self.calllinecount) + " TimeStamp: " + self.timestamp + ", XID: " + self.xid + " " + self.op + " ARGS: " + self.opdata + " RepLine: " + str(self.replylinecount) + " " + self.replydata + +class NFSLogAnalyzer: + + def __init__ (self, optn, trackfilename, tracknamefh, stats): + self.stats = stats + self.xid_request_map = {} + self.orphan_replies = {} + self.rqlist = [] + self.CALL = 1 + self.REPLY = 2 + self.optn = optn + self.trackfilename = trackfilename + self.tracknamefh = tracknamefh + self.trackedfilehandles = [] + + def handle_call_line (self, logline, linecount): + newreq = NFSRequest (logline, linecount) + xid = newreq.getXID () + if (self.optn == SYNTHESIZE): + self.xid_request_map [xid] = newreq + self.rqlist.append(newreq) + elif self.optn == TRACKFILENAME: + if newreq.requestIsEntryOp(): + if newreq.entryname == self.trackfilename: + self.xid_request_map [xid] = newreq + self.rqlist.append(newreq) + else: + del newreq + elif self.tracknamefh == ENABLE_TRACKNAME_FH: + if len (self.trackedfilehandles) > 0: + if newreq.gfid in self.trackedfilehandles: + self.xid_request_map [xid] = newreq + self.rqlist.append(newreq) + else: + del newreq + else: + del newreq + else: + del newreq + + + def handle_reply_line (self, logline, linecount): + tokens = logline.strip ().split (" ") + + xididx = tokens.index ("XID:") + xid = tokens [xididx + 1].strip(",") + if xid not in self.xid_request_map.keys (): + self.orphan_replies [xid] = logline + else: + rq = self.xid_request_map [xid] + rq.setReply (logline, linecount) + if rq.requestIsEntryOp() and rq.entryname == self.trackfilename: + self.trackedfilehandles.append (rq.replygfid) + + def analyzeLine (self, logline, linecount): + tokens = logline.strip ().split (" ") + msgtype = 0 + + if "XID:" not in tokens: + return + + if "args:" in tokens: + msgtype = self.CALL + elif "NFS:" in tokens: + msgtype = self.REPLY + + if msgtype == self.CALL: + self.handle_call_line (logline, linecount) + elif msgtype == self.REPLY: + self.handle_reply_line (logline, linecount) + + def getStats (self): + if self.stats == 0: + return + rcount = len (self.xid_request_map.keys ()) + orphancount = len (self.orphan_replies.keys ()) + print "Requests: " + str(rcount) + ", Orphans: " + str(orphancount) + + def dump (self): + self.getStats () + for rq in self.rqlist: + rq.dump () + del rq + + self.rqlist = [] + self.orphan_replies = {} + self.xid_request_map = {} + + +linecount = 0 + +SYNTHESIZE = 1 +TRACKFILENAME = 2 + +ENABLESTATS = 1 +DISABLESTATS = 0 + +ENABLE_TRACKNAME_FH = 1 +DISABLE_TRACKNAME_FH = 0 + +progmsgcount = 1000 +dumpinterval = 200000 +operation = SYNTHESIZE +stats = ENABLESTATS +tracknamefh = DISABLE_TRACKNAME_FH +trackfilename = "" + +""" +Print the progress of the analysing operations every X number of lines read from +the logs, where X is the argument provided to this option. + +Use this to print a status message every say 10000 lines processed or 100000 +lines processed to know how much longer the processing will go on for. + + +USAGE: --progress <NUMLINES> +""" +if "--progress" in sys.argv: + idx = sys.argv.index ("--progress") + progmsgcount = int(sys.argv[idx+1]) + +""" +The replies for a NFS request can be separated by hundreds and even thousands +of other NFS requests and replies. These can be spread over many hundreds and +thousands of log lines. This script maintains a memory dict to map each request +to its reply using the XID. Because this is in-core, there is a limit to the +number of entries in the dict. At regular intervals, it dumps the mapped +requests and the replies into the stdout. The requests whose replies were not +found at the point of dumping are left as orphans, i.e. without info about the +replies. Use this option to tune the number of lines to maximize the number of +requests whose replies are found while balancing the dict size with memory +on the machine. The default works fine for most cases. + +USAGE: --dump <NUMLINES> +""" +if "--dump" in sys.argv: + idx = sys.argv.index ("--dump") + dumpinterval = int(sys.argv[idx+1]) + +""" +The default operation of the script is to output all the requests mapped to +their replies in a single line. This operation mode can be changed by this +argument. It is used to print only those operations that were performed on the +filename given as the argument to this option. Only those entry operations are +printed which contain this filename. + +USAGE: --trackfilename <filename> +""" +if "--trackfilename" in sys.argv: + idx = sys.argv.index ("--trackfilename") + trackfilename = sys.argv[idx + 1] + operation = TRACKFILENAME + +""" +At every dump interval, some stats are printed about the dumped lines. +Use this option to disable printing that to avoid cluttering the +output. +""" +if "--nostats" in sys.argv: + stats = DISABLESTATS + +""" +While tracking a file using --trackfilename, we're only given those +operations which contain the filename. This excludes a large number +of operations which operate on that file using its filehandle instead of +the filename. This option enables outputting those operations also. It +tracks every single file handle that was ever seen in the log for a given +filename. + +USAGE: --trackfilename +""" +if "--tracknamefh" in sys.argv: + tracknamefh = ENABLE_TRACKNAME_FH + +la = NFSLogAnalyzer (operation, trackfilename, tracknamefh, stats) + +for line in sys.stdin: + linecount = linecount + 1 + if linecount % dumpinterval == 0: + sys.stderr.write ("Dumping data..\n") + la.dump () + + if linecount % progmsgcount == 0: + sys.stderr.write ("Integrating line: "+ str(linecount) + "\n") + la.analyzeLine (line, linecount) diff --git a/extras/group-virt.example b/extras/group-virt.example new file mode 100644 index 000000000..7dc777f2d --- /dev/null +++ b/extras/group-virt.example @@ -0,0 +1,6 @@ +quick-read=off +read-ahead=off +io-cache=off +stat-prefetch=off +eager-lock=enable +remote-dio=enable diff --git a/extras/hook-scripts/Makefile.am b/extras/hook-scripts/Makefile.am new file mode 100644 index 000000000..f6bded20c --- /dev/null +++ b/extras/hook-scripts/Makefile.am @@ -0,0 +1 @@ +EXTRA_DIST = S29CTDBsetup.sh S30samba-start.sh S30samba-stop.sh S30samba-set.sh S56glusterd-geo-rep-create-post.sh diff --git a/extras/hook-scripts/S29CTDBsetup.sh b/extras/hook-scripts/S29CTDBsetup.sh new file mode 100644 index 000000000..e256be1f3 --- /dev/null +++ b/extras/hook-scripts/S29CTDBsetup.sh @@ -0,0 +1,69 @@ +#! /bin/bash +#non-portable - RHS-2.0 only +# - The script mounts the 'meta-vol' on start 'event' on a known +# directory (eg. /gluster/lock) +# - Adds the necessary configuration changes for ctdb in smb.conf and +# restarts smb service. +# - P.S: There are other 'tasks' that need to be done outside this script +# to get CTDB based failover up and running. + +SMB_CONF=/etc/samba/smb.conf + +CTDB_MNT=/gluster/lock +PROGNAME="ctdb" +OPTSPEC="volname:" +VOL= +# $META is the volume that will be used by CTDB as a shared filesystem. +# It is not desirable to use this volume for storing 'data' as well. +# META is set to 'all' (viz. a keyword and hence not a legal volume name) +# to prevent the script from running for volumes it was not intended. +# User needs to set META to the volume that serves CTDB lockfile. +META="all" + +function sighup_samba () { + pid=`cat /var/run/smbd.pid` + if [ $pid != " " ] + then + kill -HUP $pid; + else + /etc/init.d/smb start + fi +} + +function parse_args () { + ARGS=$(getopt -l $OPTSPEC -name $PROGNAME $@) + eval set -- "$ARGS" + + while true; do + case $1 in + --volname) + shift + VOL=$1 + ;; + + *) + shift + break + ;; + + esac + + shift + done +} + +function add_glusterfs_ctdb_options () { + PAT="Share Definitions" + GLUSTER_CTDB_CONFIG="# ctdb config for glusterfs\n\tclustering = yes\n\tidmap backend = tdb2\n\tprivate dir = "$CTDB_MNT"\n" + + sed -i /"$PAT"/i\ "$GLUSTER_CTDB_CONFIG" $SMB_CONF +} + +parse_args $@ +if [ "$META" = "$VOL" ] +then + add_glusterfs_ctdb_options + sighup_samba + mount -t glusterfs `hostname`:$VOL "$CTDB_MNT" & +fi + diff --git a/extras/hook-scripts/S30samba-set.sh b/extras/hook-scripts/S30samba-set.sh new file mode 100755 index 000000000..6b11f5a4f --- /dev/null +++ b/extras/hook-scripts/S30samba-set.sh @@ -0,0 +1,109 @@ +#!/bin/bash + +#Need to be copied to hooks/<HOOKS_VER>/set/post/ + +#TODO: All gluster and samba paths are assumed for fedora like systems. +#Some efforts are required to make it work on other distros. + +#The preferred way of creating a smb share of a gluster volume has changed. +#The old method was to create a fuse mount of the volume and share the mount +#point through samba. +# +#New method eliminates the requirement of fuse mount and changes in fstab. +#glusterfs_vfs plugin for samba makes call to libgfapi to access the volume. +# +#This hook script enables user to enable or disable smb share by volume set +#option. Keys "user.cifs" and "user.smb" both are valid, but user.smb is +#preferred. + + +PROGNAME="Ssamba-set" +OPTSPEC="volname:" +VOL= + +enable_smb="" + +function parse_args () { + ARGS=$(getopt -l $OPTSPEC -o "o" -name $PROGNAME $@) + eval set -- "$ARGS" + + while true; do + case $1 in + --volname) + shift + VOL=$1 + ;; + *) + shift + for pair in $@; do + read key value < <(echo "$pair" | tr "=" " ") + case "$key" in + "user.cifs") + enable_smb=$value + ;; + "user.smb") + enable_smb=$value + ;; + *) + ;; + esac + done + + shift + break + ;; + esac + shift + done +} + +function add_samba_share () { + volname=$1 + STRING="\n[gluster-$volname]\n" + STRING+="comment = For samba share of volume $volname\n" + STRING+="vfs objects = glusterfs\n" + STRING+="glusterfs:volume = $volname\n" + STRING+="glusterfs:logfile = /var/log/samba/glusterfs-$volname.%%M.log\n" + STRING+="glusterfs:loglevel = 7\n" + STRING+="path = /\n" + STRING+="read only = no\n" + STRING+="guest ok = yes\n" + printf "$STRING" >> /etc/samba/smb.conf +} + +function sighup_samba () { + pid=`cat /var/run/smbd.pid` + if [ "x$pid" != "x" ] + then + kill -HUP "$pid"; + else + /etc/init.d/smb start + fi +} + +function del_samba_share () { + volname=$1 + sed -i "/\[gluster-$volname\]/,/^$/d" /etc/samba/smb.conf +} + +function is_volume_started () { + volname=$1 + echo "$(grep status /var/lib/glusterd/vols/"$volname"/info |\ + cut -d"=" -f2)" +} + +parse_args $@ +if [ "0" = $(is_volume_started "$VOL") ]; then + exit 0 +fi + +if [ "$enable_smb" = "enable" ]; then + if ! grep --quiet "\[gluster-$VOL\]" /etc/samba/smb.conf ; then + add_samba_share $VOL + sighup_samba + fi + +elif [ "$enable_smb" = "disable" ]; then + del_samba_share $VOL + sighup_samba +fi diff --git a/extras/hook-scripts/S30samba-start.sh b/extras/hook-scripts/S30samba-start.sh new file mode 100755 index 000000000..34fde0ef8 --- /dev/null +++ b/extras/hook-scripts/S30samba-start.sh @@ -0,0 +1,110 @@ +#!/bin/bash + +#Need to be copied to hooks/<HOOKS_VER>/start/post + +#TODO: All gluster and samba paths are assumed for fedora like systems. +#Some efforts are required to make it work on other distros. + +#The preferred way of creating a smb share of a gluster volume has changed. +#The old method was to create a fuse mount of the volume and share the mount +#point through samba. +# +#New method eliminates the requirement of fuse mount and changes in fstab. +#glusterfs_vfs plugin for samba makes call to libgfapi to access the volume. +# +#This hook script automagically creates shares for volume on every volume start +#event by adding the entries in smb.conf file and sending SIGHUP to samba. +# +#In smb.conf: +#glusterfs vfs plugin has to be specified as required vfs object. +#Path value is relative to the root of gluster volume;"/" signifies complete +#volume. + +PROGNAME="Ssamba-start" +OPTSPEC="volname:" +VOL= +CONFIGFILE= +LOGFILEBASE= +PIDDIR= + +function parse_args () { + ARGS=$(getopt -l $OPTSPEC -name $PROGNAME $@) + eval set -- "$ARGS" + + while true; do + case $1 in + --volname) + shift + VOL=$1 + ;; + *) + shift + break + ;; + esac + shift + done +} + +function find_config_info () { + cmdout=`smbd -b | grep smb.conf` + if [ $? -ne 0 ];then + echo "Samba is not installed" + exit 1 + fi + CONFIGFILE=`echo $cmdout | awk {'print $2'}` + PIDDIR=`smbd -b | grep PIDDIR | awk {'print $2'}` + LOGFILEBASE=`smbd -b | grep 'LOGFILEBASE' | awk '{print $2}'` +} + +function add_samba_share () { + volname=$1 + STRING="\n[gluster-$volname]\n" + STRING+="comment = For samba share of volume $volname\n" + STRING+="vfs objects = glusterfs\n" + STRING+="glusterfs:volume = $volname\n" + STRING+="glusterfs:logfile = $LOGFILEBASE/glusterfs-$volname.%%M.log\n" + STRING+="glusterfs:loglevel = 7\n" + STRING+="path = /\n" + STRING+="read only = no\n" + STRING+="guest ok = yes\n" + printf "$STRING" >> ${CONFIGFILE} +} + +function sighup_samba () { + pid=`cat ${PIDDIR}/smbd.pid` + if [ "x$pid" != "x" ] + then + kill -HUP "$pid"; + else + /etc/init.d/smb condrestart + fi +} + +function get_smb () { + volname=$1 + uservalue= + + usercifsvalue=$(grep user.cifs /var/lib/glusterd/vols/"$volname"/info |\ + cut -d"=" -f2) + usersmbvalue=$(grep user.smb /var/lib/glusterd/vols/"$volname"/info |\ + cut -d"=" -f2) + + if [[ $usercifsvalue = "disable" || $usersmbvalue = "disable" ]]; then + uservalue="disable" + fi + echo "$uservalue" +} + +parse_args $@ +if [ $(get_smb "$VOL") = "disable" ]; then + exit 0 +fi + +#Find smb.conf, smbd pid directory and smbd logfile path +find_config_info + +if ! grep --quiet "\[gluster-$VOL\]" ${CONFIGFILE} ; then + add_samba_share $VOL + sighup_samba +fi diff --git a/extras/hook-scripts/S30samba-stop.sh b/extras/hook-scripts/S30samba-stop.sh new file mode 100755 index 000000000..8950eea43 --- /dev/null +++ b/extras/hook-scripts/S30samba-stop.sh @@ -0,0 +1,71 @@ +#! /bin/bash + +#Need to be copied to hooks/<HOOKS_VER>/stop/pre + +#TODO: All gluster and samba paths are assumed for fedora like systems. +#Some efforts are required to make it work on other distros. + +#The preferred way of creating a smb share of a gluster volume has changed. +#The old method was to create a fuse mount of the volume and share the mount +#point through samba. +# +#New method eliminates the requirement of fuse mount and changes in fstab. +#glusterfs_vfs plugin for samba makes call to libgfapi to access the volume. +# +#This hook script automagically removes shares for volume on every volume stop +#event by removing the volume related entries(if any) in smb.conf file. + +PROGNAME="Ssamba-stop" +OPTSPEC="volname:" +VOL= +CONFIGFILE= +PIDDIR= + +function parse_args () { + ARGS=$(getopt -l $OPTSPEC -name $PROGNAME $@) + eval set -- "$ARGS" + + while true; do + case $1 in + --volname) + shift + VOL=$1 + ;; + *) + shift + break + ;; + esac + shift + done +} + +function find_config_info () { + cmdout=`smbd -b | grep smb.conf` + if [ $? -ne 0 ];then + echo "Samba is not installed" + exit 1 + fi + CONFIGFILE=`echo $cmdout | awk {'print $2'}` + PIDDIR=`smbd -b | grep PIDDIR | awk {'print $2'}` +} + +function del_samba_share () { + volname=$1 + sed -i "/\[gluster-$volname\]/,/^$/d" ${CONFIGFILE} +} + +function sighup_samba () { + pid=`cat ${PIDDIR}/smbd.pid` + if [ "x$pid" != "x" ] + then + kill -HUP $pid; + else + /etc/init.d/smb condrestart + fi +} + +parse_args $@ +find_config_info +del_samba_share $VOL +sighup_samba diff --git a/extras/hook-scripts/S40ufo-stop.py b/extras/hook-scripts/S40ufo-stop.py new file mode 100755 index 000000000..107f19683 --- /dev/null +++ b/extras/hook-scripts/S40ufo-stop.py @@ -0,0 +1,24 @@ +#!/usr/bin/env python + +import os +from optparse import OptionParser + +if __name__ == '__main__': + # check if swift is installed + try: + from gluster.swift.common.Glusterfs import get_mnt_point, unmount + except ImportError: + import sys + sys.exit("Openstack Swift does not appear to be installed properly"); + + op = OptionParser(usage="%prog [options...]") + op.add_option('--volname', dest='vol', type=str) + op.add_option('--last', dest='last', type=str) + (opts, args) = op.parse_args() + + + mnt_point = get_mnt_point(opts.vol) + if mnt_point: + unmount(mnt_point) + else: + sys.exit("get_mnt_point returned none for mount point") diff --git a/extras/hook-scripts/S56glusterd-geo-rep-create-post.sh b/extras/hook-scripts/S56glusterd-geo-rep-create-post.sh new file mode 100755 index 000000000..1369c22fc --- /dev/null +++ b/extras/hook-scripts/S56glusterd-geo-rep-create-post.sh @@ -0,0 +1,42 @@ +#!/bin/bash + +key_val_pair1=`echo $2 | cut -d ' ' -f 1` +key_val_pair2=`echo $2 | cut -d ' ' -f 2` +key_val_pair3=`echo $2 | cut -d ' ' -f 3` + +key=`echo $key_val_pair1 | cut -d '=' -f 1` +val=`echo $key_val_pair1 | cut -d '=' -f 2` +if [ "$key" != "is_push_pem" ]; then + exit; +fi +if [ "$val" != '1' ]; then + exit; +fi + +key=`echo $key_val_pair2 | cut -d '=' -f 1` +val=`echo $key_val_pair2 | cut -d '=' -f 2` +if [ "$key" != "pub_file" ]; then + exit; +fi +if [ "$val" == "" ]; then + exit; +fi +pub_file=`echo $val` +pub_file_tmp=`echo $val`_tmp + +key=`echo $key_val_pair3 | cut -d '=' -f 1` +val=`echo $key_val_pair3 | cut -d '=' -f 2` +if [ "$key" != "slave_ip" ]; then + exit; +fi +if [ "$val" == "" ]; then + exit; +fi +slave_ip=`echo $val` + +if [ -f $pub_file ]; then + scp $pub_file $slave_ip:$pub_file_tmp + ssh $slave_ip "mv $pub_file_tmp $pub_file" + ssh $slave_ip "gluster system:: copy file /geo-replication/common_secret.pem.pub > /dev/null" + ssh $slave_ip "gluster system:: execute add_secret_pub > /dev/null" +fi diff --git a/extras/init.d/Makefile.am b/extras/init.d/Makefile.am index 7d8deb74f..38898fddd 100644 --- a/extras/init.d/Makefile.am +++ b/extras/init.d/Makefile.am @@ -1,17 +1,22 @@ -EXTRA_DIST = glusterfsd-Debian glusterfsd-Redhat glusterfsd-SuSE glusterfs-server.plist +EXTRA_DIST = glusterd-Debian glusterd-Redhat glusterd-SuSE glusterd.plist rhel5-load-fuse.modules CLEANFILES = -initdir = /etc/init.d +INIT_DIR = @initdir@ +SYSTEMD_DIR = @systemddir@ +LAUNCHD_DIR = @launchddir@ $(GF_DISTRIBUTION): - $(mkdir_p) $(DESTDIR)$(initdir) - $(INSTALL_PROGRAM) $(top_srcdir)/extras/init.d/glusterfsd-$(GF_DISTRIBUTION) $(DESTDIR)$(initdir)/glusterfsd + @if [ ! -d $(SYSTEMD_DIR) ]; then \ + $(mkdir_p) $(DESTDIR)$(INIT_DIR); \ + $(INSTALL_PROGRAM) glusterd-$(GF_DISTRIBUTION) $(DESTDIR)$(INIT_DIR)/glusterd; \ + fi install-exec-local: $(GF_DISTRIBUTION) install-data-local: if GF_DARWIN_HOST_OS - cp glusterfs-server.plist /Library/LaunchDaemons/com.zresearch.glusterfs.plist + $(mkdir_p) $(DESTDIR)$(LAUNCHD_DIR) + $(INSTALL_PROGRAM) glusterd.plist $(DESTDIR)$(LAUNCHD_DIR)/com.gluster.glusterd.plist endif diff --git a/extras/init.d/glusterfsd-Debian b/extras/init.d/glusterd-Debian.in index dcd3e0c50..918f8592c 100755 --- a/extras/init.d/glusterfsd-Debian +++ b/extras/init.d/glusterd-Debian.in @@ -1,24 +1,23 @@ #!/bin/sh ### BEGIN INIT INFO -# Provides: glusterfsd +# Provides: glusterd # Required-Start: $local_fs $network # Required-Stop: $local_fs $network # Default-Start: 2 3 4 5 # Default-Stop: 0 1 6 -# Short-Description: gluster server -# Description: This file starts / stops the gluster server +# Short-Description: Gluster File System service for volume management +# Description: Gluster File System service for volume management ### END INIT INFO # Author: Chris AtLee <chris@atlee.ca> # Patched by: Matthias Albert < matthias@linux4experts.de> PATH=/sbin:/usr/sbin:/bin:/usr/bin -NAME=glusterfsd +NAME=glusterd SCRIPTNAME=/etc/init.d/$NAME -DAEMON=/usr/sbin/$NAME +DAEMON=@prefix@/sbin/$NAME PIDFILE=/var/run/$NAME.pid -CONFIGFILE=/etc/glusterfs/glusterfsd.vol -GLUSTERFS_OPTS="-f $CONFIGFILE" +GLUSTERD_OPTS="" PID=`test -f $PIDFILE && cat $PIDFILE` @@ -31,24 +30,16 @@ test -x $DAEMON || exit 0 # Define LSB log_* functions. . /lib/lsb/init-functions -check_config() -{ - if [ ! -f "$CONFIGFILE" ]; then - echo "Config file $CONFIGFILE is missing...exiting!" - exit 0 - fi -} do_start() { - check_config; pidofproc -p $PIDFILE $DAEMON >/dev/null status=$? if [ $status -eq 0 ]; then - log_success_msg "glusterfs server is already running with pid $PID" + log_success_msg "glusterd service is already running with pid $PID" else - log_daemon_msg "Starting glusterfs server" "glusterfsd" - start-stop-daemon --start --quiet --oknodo --pidfile $PIDFILE --startas $DAEMON -- -p $PIDFILE $GLUSTERFS_OPTS + log_daemon_msg "Starting glusterd service" "glusterd" + start-stop-daemon --start --quiet --oknodo --pidfile $PIDFILE --startas $DAEMON -- -p $PIDFILE $GLUSTERD_OPTS log_end_msg $? start_daemon -p $PIDFILE $DAEMON -f $CONFIGFILE return $? @@ -57,7 +48,7 @@ do_start() do_stop() { - log_daemon_msg "Stopping glusterfs server" "glusterfsd" + log_daemon_msg "Stopping glusterd service" "glusterd" start-stop-daemon --stop --quiet --oknodo --pidfile $PIDFILE log_end_msg $? rm -f $PIDFILE @@ -70,9 +61,9 @@ do_status() pidofproc -p $PIDFILE $DAEMON >/dev/null status=$? if [ $status -eq 0 ]; then - log_success_msg "glusterfs server is running with pid $PID" + log_success_msg "glusterd service is running with pid $PID" else - log_failure_msg "glusterfs server is not running." + log_failure_msg "glusterd service is not running." fi exit $status } diff --git a/extras/init.d/glusterd-Redhat.in b/extras/init.d/glusterd-Redhat.in new file mode 100755 index 000000000..e320708ae --- /dev/null +++ b/extras/init.d/glusterd-Redhat.in @@ -0,0 +1,142 @@ +#!/bin/bash +# +# glusterd Startup script for the glusterfs server +# +# chkconfig: - 20 80 +# description: Clustered file-system server + +### BEGIN INIT INFO +# Provides: glusterd +# Required-Start: $local_fs $network +# Required-Stop: $local_fs $network +# Should-Start: +# Should-Stop: +# Default-Start: +# Default-Stop: 0 1 2 3 4 5 6 +# Short-Description: glusterfs server +# Description: Clustered file-system server +### END INIT INFO +# + +# Source function library. +. /etc/rc.d/init.d/functions + +BASE=glusterd + +# Fedora File System Layout dictates /run +[ -e /run ] && RUNDIR="/run" +PIDFILE="${RUNDIR:-/var/run}/${BASE}.pid" + +PID=`test -f $PIDFILE && cat $PIDFILE` + +# Overwriteable from sysconfig +LOG_LEVEL='' +LOG_FILE='' +GLUSTERD_OPTIONS='' +GLUSTERD_NOFILE='65536' + +[ -f /etc/sysconfig/${BASE} ] && . /etc/sysconfig/${BASE} + +[ ! -z $LOG_LEVEL ] && GLUSTERD_OPTIONS="${GLUSTERD_OPTIONS} --log-level ${LOG_LEVEL}" +[ ! -z $LOG_FILE ] && GLUSTERD_OPTIONS="${GLUSTERD_OPTIONS} --log-file ${LOG_FILE}" + +GLUSTERFSD=glusterfsd +GLUSTERFS=glusterfs +GLUSTERD_BIN=@prefix@/sbin/$BASE +GLUSTERD_OPTS="--pid-file=$PIDFILE ${GLUSTERD_OPTIONS}" +GLUSTERD="$GLUSTERD_BIN $GLUSTERD_OPTS" +RETVAL=0 + +LOCKFILE=/var/lock/subsys/${BASE} + +# Start the service $BASE +start() +{ + if pidofproc -p $PIDFILE $GLUSTERD_BIN &> /dev/null; then + echo "glusterd service is already running with pid $PID" + return 0 + else + ulimit -n $GLUSTERD_NOFILE + echo -n $"Starting $BASE:" + daemon $GLUSTERD + RETVAL=$? + echo + [ $RETVAL -eq 0 ] && touch $LOCKFILE + return $RETVAL + fi +} + +# Stop the service $BASE +stop() +{ + echo -n $"Stopping $BASE:" + if pidofproc -p $PIDFILE $GLUSTERD_BIN &> /dev/null; then + killproc -p $PIDFILE $BASE + else + killproc $BASE + fi + RETVAL=$? + [ $RETVAL -eq 0 ] && rm -f $LOCKFILE + return $RETVAL +} + +restart() +{ + stop + start +} + +reload() +{ + restart +} + +force_reload() +{ + restart +} + +rh_status() +{ + status $BASE +} + +rh_status_q() +{ + rh_status &>/dev/null +} + + +### service arguments ### +case $1 in + start) + rh_status_q && exit 0 + $1 + ;; + stop) + rh_status_q || exit 0 + $1 + ;; + restart) + $1 + ;; + reload) + rh_status_q || exit 7 + $1 + ;; + force-reload) + force_reload + ;; + status) + rh_status + ;; + condrestart|try-restart) + rh_status_q || exit 0 + restart + ;; + *) + echo $"Usage: $0 {start|stop|status|restart|condrestart|try-restart|reload|force-reload}" + exit 1 +esac + +exit $? diff --git a/extras/init.d/glusterfsd-SuSE b/extras/init.d/glusterd-SuSE.in index ede27b2b3..6259bab00 100755 --- a/extras/init.d/glusterfsd-SuSE +++ b/extras/init.d/glusterd-SuSE.in @@ -1,28 +1,30 @@ #!/bin/bash # ### BEGIN INIT INFO -# Provides: glusterfsd -# Required-Start: $local_fs $network -# Required-Stop: +# Provides: glusterd +# Required-Start: $remote_fs $network +# Required-Stop: $remote_fs $network # Default-Start: 3 5 # Default-Stop: -# Short-Description: GlusterFS server daemon -# Description: All necessary services for GlusterFS clients +# Short-Description: Gluster File System service for volume management +# Description: Gluster File System service for volume management ### END INIT INFO # Get function from functions library . /etc/rc.status -BASE=glusterfsd -GSERVER="/usr/sbin/$BASE -f /etc/glusterfs/glusterfsd.vol" +BASE=glusterd +GLUSTERD_BIN=@prefix@/sbin/$BASE +GLUSTERD_OPTS="" +GLUSTERD="$GLUSTERD_BIN $GLUSTERD_OPTS" RETVAL=0 # Start the service $BASE start() { echo -n $"Starting $BASE:" - startproc $GSERVER + startproc $GLUSTERD return $? } @@ -30,7 +32,7 @@ start() stop() { echo $"Stopping $BASE:" - killproc $BASE + killproc $BASE return $? } @@ -52,20 +54,24 @@ case $1 in rc_status -v ;; status) - echo -n " glusterfsd" + echo -n " glusterd" if ! checkproc $BASE ;then echo " not running" rc_failed 3 fi rc_status -v ;; + reload) + rc_failed 3 + rc_status -v + ;; restart) $0 stop $0 start rc_status ;; *) - echo $"Usage: $0 {start|stop|status|restart}." + echo $"Usage: $0 {start|stop|status|reload|restart}." exit 1 esac diff --git a/extras/init.d/glusterfs-server.plist.in b/extras/init.d/glusterd.plist.in index 4d2287c57..7385fa486 100644 --- a/extras/init.d/glusterfs-server.plist.in +++ b/extras/init.d/glusterd.plist.in @@ -3,13 +3,11 @@ <plist version="1.0"> <dict> <key>Label</key> - <string>com.zresearch.glusterfs</string> + <string>com.gluster.glusterd</string> <key>ProgramArguments</key> <array> - <string>@prefix@/sbin/glusterfsd</string> + <string>@prefix@/sbin/glusterd</string> <string>-N</string> - <string>-f</string> - <string>@prefix@/etc/glusterfs/server.vol</string> </array> </dict> </plist> diff --git a/extras/init.d/glusterfsd-Redhat b/extras/init.d/glusterfsd-Redhat deleted file mode 100755 index 2577e9e76..000000000 --- a/extras/init.d/glusterfsd-Redhat +++ /dev/null @@ -1,51 +0,0 @@ -#!/bin/bash -# -# chkconfig: 35 90 12 -# description: Glusterfsd server -# - -# Get function from functions library -. /etc/rc.d/init.d/functions - -BASE=glusterfsd -GSERVER="/usr/sbin/$BASE -f /etc/glusterfs/glusterfsd.vol" -RETVAL=0 - -# Start the service $BASE -start() -{ - echo $"Starting $BASE:" - daemon $GSERVER - RETVAL=$? - [ $RETVAL -ne 0 ] && exit $RETVAL -} - -# Stop the service $BASE -stop() -{ - echo $"Stopping $BASE:" - killproc $BASE -} - - -### service arguments ### -case $1 in - start) - start - ;; - stop) - stop - ;; - status) - status $BASE - ;; - restart) - $0 stop - $0 start - ;; - *) - echo $"Usage: $0 {start|stop|status|restart}." - exit 1 -esac - -exit 0 diff --git a/extras/init.d/rhel5-load-fuse.modules b/extras/init.d/rhel5-load-fuse.modules new file mode 100755 index 000000000..ee194db99 --- /dev/null +++ b/extras/init.d/rhel5-load-fuse.modules @@ -0,0 +1,7 @@ +#!/bin/sh +# +# fusermount-glusterfs requires the /dev/fuse character device. The fuse module +# provides this and is loaded on demand in newer Linux distributions. +# + +[ -c /dev/fuse ] || /sbin/modprobe fuse diff --git a/extras/logger.conf.example b/extras/logger.conf.example new file mode 100644 index 000000000..248be5bda --- /dev/null +++ b/extras/logger.conf.example @@ -0,0 +1,13 @@ +# +# Sample logger.conf file to configure enhanced Logging in GlusterFS +# +# To enable enhanced logging capabilities, +# +# 1. rename this file to /etc/glusterfs/logger.conf +# +# 2. rename /etc/rsyslog.d/gluster.conf.example to +# /etc/rsyslog.d/gluster.conf +# +# This change requires restart of all gluster services/volumes and +# rsyslog. +# diff --git a/extras/ocf/Makefile.am b/extras/ocf/Makefile.am new file mode 100644 index 000000000..c49a835fb --- /dev/null +++ b/extras/ocf/Makefile.am @@ -0,0 +1,11 @@ +EXTRA_DIST = glusterd.in volume.in + +# The root of the OCF resource agent hierarchy +# Per the OCF standard, it's always "lib", +# not "lib64" (even on 64-bit platforms). +ocfdir = $(prefix)/lib/ocf + +# The ceph provider directory +radir = $(ocfdir)/resource.d/$(PACKAGE_NAME) + +ra_SCRIPTS = glusterd volume diff --git a/extras/ocf/glusterd.in b/extras/ocf/glusterd.in new file mode 100755 index 000000000..c119a285d --- /dev/null +++ b/extras/ocf/glusterd.in @@ -0,0 +1,212 @@ +#!/bin/sh +# +# glusterd +# +# Description: Manages a glusterd server as a (typically cloned) +# HA resource +# +# Authors: Florian Haas (hastexo Professional Services GmbH) +# +# License: GNU General Public License (GPL) + +####################################################################### +# Initialization: + +: ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat} +. ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs + +# Convenience variables +# When sysconfdir and localstatedir aren't passed in as +# configure flags, they're defined in terms of prefix +prefix=@prefix@ +####################################################################### + + +OCF_RESKEY_binary_default="glusterd" +OCF_RESKEY_pid_default="@localstatedir@/run/glusterd.pid" +OCF_RESKEY_socket_default="" +OCF_RESKEY_additional_parameters_default="" + +: ${OCF_RESKEY_binary=${OCF_RESKEY_binary_default}} +: ${OCF_RESKEY_pid=${OCF_RESKEY_pid_default}} + +glusterd_meta_data() { + cat <<EOF +<?xml version="1.0"?> +<!DOCTYPE resource-agent SYSTEM "ra-api-1.dtd"> +<resource-agent name="glusterd" version="0.1"> + <version>0.1</version> + <longdesc lang="en"> + </longdesc> + <shortdesc lang="en">Manages a Gluster server</shortdesc> + <parameters> + <parameter name="binary"> + <longdesc lang="en"> + Name of the glusterd executable. Specify a full absolute + path if the binary is not in your \$PATH. + </longdesc> + <shortdesc lang="en">glusterd executable</shortdesc> + <content type="string" default="$OCF_RESKEY_binary_default"/> + </parameter> + <parameter name="pid"> + <longdesc lang="en"> + Path to the glusterd PID file. + </longdesc> + <shortdesc lang="en">PID file</shortdesc> + <content type="string" default="$OCF_RESKEY_pid_default"/> + </parameter> + <parameter name="socket"> + <longdesc lang="en"> + Path to the glusterd UNIX socket file. If unspecified, + glusterd will not listen on any socket. + </longdesc> + <shortdesc lang="en">Socket file</shortdesc> + <content type="string"/> + </parameter> + </parameters> + <actions> + <action name="start" timeout="20" /> + <action name="stop" timeout="20" /> + <action name="monitor" timeout="20" interval="10" /> + <action name="reload" timeout="20" /> + <action name="meta-data" timeout="5" /> + <action name="validate-all" timeout="20" /> + </actions> +</resource-agent> +EOF + +} + +glusterd_start() { + local glusterd_options + # exit immediately if configuration is not valid + glusterd_validate_all || exit $? + + # if resource is already running, bail out early + if glusterd_monitor; then + ocf_log info "Resource is already running" + return $OCF_SUCCESS + fi + + # actually start up the resource here (make sure to immediately + # exit with an $OCF_ERR_ error code if anything goes seriously + # wrong) + glusterd_options="-p $OCF_RESKEY_pid" + if [ -n "$OCF_RESKEY_socket" ]; then + glusterd_options="$glusterd_options -S $OCF_RESKEY_socket" + fi + if [ -n "$OCF_RESKEY_additional_parameters" ]; then + glusterd_options="$glusterd_options $OCF_RESKEY_additional_parameters" + fi + + ocf_run $OCF_RESKEY_binary $glusterd_options || exit $OCF_ERR_GENERIC + + # After the resource has been started, check whether it started up + # correctly. If the resource starts asynchronously, the agent may + # spin on the monitor function here -- if the resource does not + # start up within the defined timeout, the cluster manager will + # consider the start action failed + while ! glusterd_monitor; do + ocf_log debug "Resource has not started yet, waiting" + sleep 1 + done + + # only return $OCF_SUCCESS if _everything_ succeeded as expected + return $OCF_SUCCESS +} + +glusterd_stop() { + local rc + local pid + + # exit immediately if configuration is not valid + glusterd_validate_all || exit $? + + glusterd_monitor + rc=$? + case "$rc" in + "$OCF_SUCCESS") + # Currently running. Normal, expected behavior. + ocf_log debug "Resource is currently running" + ;; + "$OCF_NOT_RUNNING") + # Currently not running. Nothing to do. + ocf_log info "Resource is already stopped" + return $OCF_SUCCESS + ;; + esac + + # actually shut down the resource here (make sure to immediately + # exit with an $OCF_ERR_ error code if anything goes seriously + # wrong) + pid=`cat $OCF_RESKEY_pid` + ocf_run kill -s TERM $pid || exit OCF_ERR_GENERIC + + # After the resource has been stopped, check whether it shut down + # correctly. If the resource stops asynchronously, the agent may + # spin on the monitor function here -- if the resource does not + # shut down within the defined timeout, the cluster manager will + # consider the stop action failed + while glusterd_monitor; do + ocf_log debug "Resource has not stopped yet, waiting" + sleep 1 + done + + # only return $OCF_SUCCESS if _everything_ succeeded as expected + return $OCF_SUCCESS + +} + +glusterd_monitor() { + local pid + + [ -e $OCF_RESKEY_pid ] || return $OCF_NOT_RUNNING + + pid=`cat $OCF_RESKEY_pid` + ocf_run kill -s 0 $pid || return $OCF_NOT_RUNNING + + ocf_log debug "$OCF_RESKEY_binary running with PID $pid" + return $OCF_SUCCESS +} + +glusterd_validate_all() { + # Test for required binaries + check_binary $OCF_RESKEY_binary + + return $OCF_SUCCESS +} + + + +# Make sure meta-data and usage always succeed +case $__OCF_ACTION in +meta-data) glusterd_meta_data + exit $OCF_SUCCESS + ;; +usage|help) glusterd_usage + exit $OCF_SUCCESS + ;; +esac + +# Anything other than meta-data and usage must pass validation +glusterd_validate_all || exit $? + +# Translate each action into the appropriate function call +case $__OCF_ACTION in +start) glusterd_start;; +stop) glusterd_stop;; +status|monitor) glusterd_monitor;; +reload) ocf_log info "Reloading..." + glusterd_start + ;; +validate-all) ;; +notify) exit $OCF_SUCCESS;; +*) glusterd_usage + exit $OCF_ERR_UNIMPLEMENTED + ;; +esac +rc=$? + +# The resource agent may optionally log a debug message +ocf_log debug "${OCF_RESOURCE_INSTANCE} $__OCF_ACTION returned $rc" +exit $rc diff --git a/extras/ocf/volume.in b/extras/ocf/volume.in new file mode 100755 index 000000000..72fd1213a --- /dev/null +++ b/extras/ocf/volume.in @@ -0,0 +1,246 @@ +#!/bin/sh +# +# glusterd +# +# Description: Manages a glusterd server as a (typically cloned) +# HA resource +# +# Authors: Florian Haas (hastexo Professional Services GmbH) +# +# License: GNU General Public License (GPL) + +####################################################################### +# Initialization: + +: ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat} +. ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs + +# Convenience variables +# When sysconfdir and localstatedir aren't passed in as +# configure flags, they're defined in terms of prefix +prefix=@prefix@ +SHORTHOSTNAME=`hostname -s` +####################################################################### + +OCF_RESKEY_binary_default="gluster" + +: ${OCF_RESKEY_binary=${OCF_RESKEY_binary_default}} + +volume_meta_data() { + cat <<EOF +<?xml version="1.0"?> +<!DOCTYPE resource-agent SYSTEM "ra-api-1.dtd"> +<resource-agent name="volume" version="0.1"> + <version>0.1</version> + <longdesc lang="en"> +Manages a GlusterFS volume and monitors its bricks. When a resource of +this type is configured as a clone (as is commonly the case), then it +must have clone ordering enabled. + </longdesc> + <shortdesc lang="en">Manages a GlusterFS volume</shortdesc> + <parameters> + <parameter name="volname" required="1"> + <longdesc lang="en"> + The name of the volume to manage. + </longdesc> + <shortdesc lang="en">volume name</shortdesc> + <content type="string"/> + </parameter> + <parameter name="binary"> + <longdesc lang="en"> + Name of the gluster executable. Specify a full absolute + path if the binary is not in your \$PATH. + </longdesc> + <shortdesc lang="en">gluster executable</shortdesc> + <content type="string" default="$OCF_RESKEY_binary_default"/> + </parameter> + </parameters> + <actions> + <action name="start" timeout="20" /> + <action name="stop" timeout="20" /> + <action name="monitor" timeout="20" interval="10" /> + <action name="reload" timeout="20" /> + <action name="meta-data" timeout="5" /> + <action name="validate-all" timeout="20" /> + </actions> +</resource-agent> +EOF + +} + +volume_getdir() { + local voldir + voldir="@sysconfdir@/glusterd/vols/${OCF_RESKEY_volname}" + + [ -d ${voldir} ] || return 1 + + echo "${voldir}" + return 0 +} + +volume_getbricks() { + local infofile + local voldir + voldir=`volume_getdir` + infofile="${voldir}/info" + + [ -e ${infofile} ] || return 1 + + echo "`sed -n -e "s/^brick-.\+=${SHORTHOSTNAME}://p" < ${infofile}`" + return 0 +} + +volume_getpids() { + local bricks + local piddir + local pidfile + local infofile + local voldir + + voldir=`volume_getdir` + bricks=`volume_getbricks` + piddir="${voldir}/run" + + for brick in ${bricks}; do + pidfile="${piddir}/${SHORTHOSTNAME}${brick}.pid" + [ -e $pidfile ] || return 1 + cat $pidfile + done + + return 0 +} + +volume_start() { + local volume_options + + # exit immediately if configuration is not valid + volume_validate_all || exit $? + + # if resource is already running, bail out early + if volume_monitor; then + ocf_log info "Resource is already running" + return $OCF_SUCCESS + fi + + # actually start up the resource here + ocf_run "$OCF_RESKEY_binary" \ + volume start "$OCF_RESKEY_volname" force || exit $OCF_ERR_GENERIC + + # After the resource has been started, check whether it started up + # correctly. If the resource starts asynchronously, the agent may + # spin on the monitor function here -- if the resource does not + # start up within the defined timeout, the cluster manager will + # consider the start action failed + while ! volume_monitor; do + ocf_log debug "Resource has not started yet, waiting" + sleep 1 + done + + # only return $OCF_SUCCESS if _everything_ succeeded as expected + return $OCF_SUCCESS +} + +volume_stop() { + local rc + local pid + + # exit immediately if configuration is not valid + volume_validate_all || exit $? + + volume_monitor + rc=$? + case "$rc" in + "$OCF_SUCCESS") + # Currently running. Normal, expected behavior. + ocf_log debug "Resource is currently running" + ;; + "$OCF_NOT_RUNNING") + # Currently not running. Nothing to do. + ocf_log info "Resource is already stopped" + return $OCF_SUCCESS + ;; + esac + + # actually shut down the resource here (make sure to immediately + # exit with an $OCF_ERR_ error code if anything goes seriously + # wrong) + pids=`volume_getpids` + for pid in $pids; do + ocf_run kill -s TERM $pid + done + + # After the resource has been stopped, check whether it shut down + # correctly. If the resource stops asynchronously, the agent may + # spin on the monitor function here -- if the resource does not + # shut down within the defined timeout, the cluster manager will + # consider the stop action failed + while volume_monitor; do + ocf_log debug "Resource has not stopped yet, waiting" + sleep 1 + done + + # only return $OCF_SUCCESS if _everything_ succeeded as expected + return $OCF_SUCCESS + +} + +volume_monitor() { + local pid + + pids=`volume_getpids` || return $OCF_NOT_RUNNING + + for pid in $pids; do + ocf_run kill -s 0 $pid || return $OCF_NOT_RUNNING + done + + ocf_log debug "Local bricks for volume ${OCF_RESKEY_volname} running with PIDs $pids" + return $OCF_SUCCESS +} + +volume_validate_all() { + # Test for configuration errors first + if [ -z "${OCF_RESKEY_volname}" ]; then + ocf_log err 'Missing required parameter "volname"' + return $OCF_ERR_CONFIGURED + fi + + # Test for required binaries + check_binary $OCF_RESKEY_binary + + return $OCF_SUCCESS +} + + + +# Make sure meta-data and usage always succeed +case $__OCF_ACTION in +meta-data) volume_meta_data + exit $OCF_SUCCESS + ;; +usage|help) volume_usage + exit $OCF_SUCCESS + ;; +esac + +# Anything other than meta-data and usage must pass validation +volume_validate_all || exit $? + +# Translate each action into the appropriate function call +case $__OCF_ACTION in +start) volume_start;; +stop) volume_stop;; +status|monitor) volume_monitor;; +reload) ocf_log info "Reloading..." + volume_start + ;; +validate-all) ;; +notify) exit $OCF_SUCCESS;; +*) volume_usage + exit $OCF_ERR_UNIMPLEMENTED + ;; +esac +rc=$? + +# The resource agent may optionally log a debug message +ocf_log debug "${OCF_RESOURCE_INSTANCE} $__OCF_ACTION returned $rc" +exit $rc diff --git a/extras/profiler/glusterfs-profiler b/extras/profiler/glusterfs-profiler new file mode 100755 index 000000000..65d445864 --- /dev/null +++ b/extras/profiler/glusterfs-profiler @@ -0,0 +1,817 @@ +#!/usr/bin/env python + +# Copyright (c) 2006-2012 Red Hat, Inc. <http://www.redhat.com> +# This file is part of GlusterFS. +# +# This file is licensed to you under your choice of the GNU Lesser +# General Public License, version 3 or any later version (LGPLv3 or +# later), or the GNU General Public License, version 2 (GPLv2), in all +# cases as published by the Free Software Foundation. + + +# texttable - module for creating simple ASCII tables +# Incorporated from texttable.py downloaded from +# http://jefke.free.fr/stuff/python/texttable/texttable-0.7.0.tar.gz + +import sys +import string + +try: + if sys.version >= '2.3': + import textwrap + elif sys.version >= '2.2': + from optparse import textwrap + else: + from optik import textwrap +except ImportError: + sys.stderr.write("Can't import textwrap module!\n") + raise + +try: + True, False +except NameError: + (True, False) = (1, 0) + +def len(iterable): + """Redefining len here so it will be able to work with non-ASCII characters + """ + if not isinstance(iterable, str): + return iterable.__len__() + + try: + return len(unicode(iterable, 'utf')) + except: + return iterable.__len__() + +class ArraySizeError(Exception): + """Exception raised when specified rows don't fit the required size + """ + + def __init__(self, msg): + self.msg = msg + Exception.__init__(self, msg, '') + + def __str__(self): + return self.msg + +class Texttable: + + BORDER = 1 + HEADER = 1 << 1 + HLINES = 1 << 2 + VLINES = 1 << 3 + + def __init__(self, max_width=80): + """Constructor + + - max_width is an integer, specifying the maximum width of the table + - if set to 0, size is unlimited, therefore cells won't be wrapped + """ + + if max_width <= 0: + max_width = False + self._max_width = max_width + self._deco = Texttable.VLINES | Texttable.HLINES | Texttable.BORDER | \ + Texttable.HEADER + self.set_chars(['-', '|', '+', '=']) + self.reset() + + def reset(self): + """Reset the instance + + - reset rows and header + """ + + self._hline_string = None + self._row_size = None + self._header = [] + self._rows = [] + + def header(self, array): + """Specify the header of the table + """ + + self._check_row_size(array) + self._header = map(str, array) + + def add_row(self, array): + """Add a row in the rows stack + + - cells can contain newlines and tabs + """ + + self._check_row_size(array) + self._rows.append(map(str, array)) + + def add_rows(self, rows, header=True): + """Add several rows in the rows stack + + - The 'rows' argument can be either an iterator returning arrays, + or a by-dimensional array + - 'header' specifies if the first row should be used as the header + of the table + """ + + # nb: don't use 'iter' on by-dimensional arrays, to get a + # usable code for python 2.1 + if header: + if hasattr(rows, '__iter__') and hasattr(rows, 'next'): + self.header(rows.next()) + else: + self.header(rows[0]) + rows = rows[1:] + for row in rows: + self.add_row(row) + + def set_chars(self, array): + """Set the characters used to draw lines between rows and columns + + - the array should contain 4 fields: + + [horizontal, vertical, corner, header] + + - default is set to: + + ['-', '|', '+', '='] + """ + + if len(array) != 4: + raise ArraySizeError, "array should contain 4 characters" + array = [ x[:1] for x in [ str(s) for s in array ] ] + (self._char_horiz, self._char_vert, + self._char_corner, self._char_header) = array + + def set_deco(self, deco): + """Set the table decoration + + - 'deco' can be a combinaison of: + + Texttable.BORDER: Border around the table + Texttable.HEADER: Horizontal line below the header + Texttable.HLINES: Horizontal lines between rows + Texttable.VLINES: Vertical lines between columns + + All of them are enabled by default + + - example: + + Texttable.BORDER | Texttable.HEADER + """ + + self._deco = deco + + def set_cols_align(self, array): + """Set the desired columns alignment + + - the elements of the array should be either "l", "c" or "r": + + * "l": column flushed left + * "c": column centered + * "r": column flushed right + """ + + self._check_row_size(array) + self._align = array + + def set_cols_valign(self, array): + """Set the desired columns vertical alignment + + - the elements of the array should be either "t", "m" or "b": + + * "t": column aligned on the top of the cell + * "m": column aligned on the middle of the cell + * "b": column aligned on the bottom of the cell + """ + + self._check_row_size(array) + self._valign = array + + def set_cols_width(self, array): + """Set the desired columns width + + - the elements of the array should be integers, specifying the + width of each column. For example: + + [10, 20, 5] + """ + + self._check_row_size(array) + try: + array = map(int, array) + if reduce(min, array) <= 0: + raise ValueError + except ValueError: + sys.stderr.write("Wrong argument in column width specification\n") + raise + self._width = array + + def draw(self): + """Draw the table + + - the table is returned as a whole string + """ + + if not self._header and not self._rows: + return + self._compute_cols_width() + self._check_align() + out = "" + if self._has_border(): + out += self._hline() + if self._header: + out += self._draw_line(self._header, isheader=True) + if self._has_header(): + out += self._hline_header() + length = 0 + for row in self._rows: + length += 1 + out += self._draw_line(row) + if self._has_hlines() and length < len(self._rows): + out += self._hline() + if self._has_border(): + out += self._hline() + return out[:-1] + + def _check_row_size(self, array): + """Check that the specified array fits the previous rows size + """ + + if not self._row_size: + self._row_size = len(array) + elif self._row_size != len(array): + raise ArraySizeError, "array should contain %d elements" \ + % self._row_size + + def _has_vlines(self): + """Return a boolean, if vlines are required or not + """ + + return self._deco & Texttable.VLINES > 0 + + def _has_hlines(self): + """Return a boolean, if hlines are required or not + """ + + return self._deco & Texttable.HLINES > 0 + + def _has_border(self): + """Return a boolean, if border is required or not + """ + + return self._deco & Texttable.BORDER > 0 + + def _has_header(self): + """Return a boolean, if header line is required or not + """ + + return self._deco & Texttable.HEADER > 0 + + def _hline_header(self): + """Print header's horizontal line + """ + + return self._build_hline(True) + + def _hline(self): + """Print an horizontal line + """ + + if not self._hline_string: + self._hline_string = self._build_hline() + return self._hline_string + + def _build_hline(self, is_header=False): + """Return a string used to separated rows or separate header from + rows + """ + horiz = self._char_horiz + if (is_header): + horiz = self._char_header + # compute cell separator + s = "%s%s%s" % (horiz, [horiz, self._char_corner][self._has_vlines()], + horiz) + # build the line + l = string.join([horiz*n for n in self._width], s) + # add border if needed + if self._has_border(): + l = "%s%s%s%s%s\n" % (self._char_corner, horiz, l, horiz, + self._char_corner) + else: + l += "\n" + return l + + def _len_cell(self, cell): + """Return the width of the cell + + Special characters are taken into account to return the width of the + cell, such like newlines and tabs + """ + + cell_lines = cell.split('\n') + maxi = 0 + for line in cell_lines: + length = 0 + parts = line.split('\t') + for part, i in zip(parts, range(1, len(parts) + 1)): + length = length + len(part) + if i < len(parts): + length = (length/8 + 1)*8 + maxi = max(maxi, length) + return maxi + + def _compute_cols_width(self): + """Return an array with the width of each column + + If a specific width has been specified, exit. If the total of the + columns width exceed the table desired width, another width will be + computed to fit, and cells will be wrapped. + """ + + if hasattr(self, "_width"): + return + maxi = [] + if self._header: + maxi = [ self._len_cell(x) for x in self._header ] + for row in self._rows: + for cell,i in zip(row, range(len(row))): + try: + maxi[i] = max(maxi[i], self._len_cell(cell)) + except (TypeError, IndexError): + maxi.append(self._len_cell(cell)) + items = len(maxi) + length = reduce(lambda x,y: x+y, maxi) + if self._max_width and length + items*3 + 1 > self._max_width: + maxi = [(self._max_width - items*3 -1) / items \ + for n in range(items)] + self._width = maxi + + def _check_align(self): + """Check if alignment has been specified, set default one if not + """ + + if not hasattr(self, "_align"): + self._align = ["l"]*self._row_size + if not hasattr(self, "_valign"): + self._valign = ["t"]*self._row_size + + def _draw_line(self, line, isheader=False): + """Draw a line + + Loop over a single cell length, over all the cells + """ + + line = self._splitit(line, isheader) + space = " " + out = "" + for i in range(len(line[0])): + if self._has_border(): + out += "%s " % self._char_vert + length = 0 + for cell, width, align in zip(line, self._width, self._align): + length += 1 + cell_line = cell[i] + fill = width - len(cell_line) + if isheader: + align = "c" + if align == "r": + out += "%s " % (fill * space + cell_line) + elif align == "c": + out += "%s " % (fill/2 * space + cell_line \ + + (fill/2 + fill%2) * space) + else: + out += "%s " % (cell_line + fill * space) + if length < len(line): + out += "%s " % [space, self._char_vert][self._has_vlines()] + out += "%s\n" % ['', self._char_vert][self._has_border()] + return out + + def _splitit(self, line, isheader): + """Split each element of line to fit the column width + + Each element is turned into a list, result of the wrapping of the + string to the desired width + """ + + line_wrapped = [] + for cell, width in zip(line, self._width): + array = [] + for c in cell.split('\n'): + array.extend(textwrap.wrap(unicode(c, 'utf'), width)) + line_wrapped.append(array) + max_cell_lines = reduce(max, map(len, line_wrapped)) + for cell, valign in zip(line_wrapped, self._valign): + if isheader: + valign = "t" + if valign == "m": + missing = max_cell_lines - len(cell) + cell[:0] = [""] * (missing / 2) + cell.extend([""] * (missing / 2 + missing % 2)) + elif valign == "b": + cell[:0] = [""] * (max_cell_lines - len(cell)) + else: + cell.extend([""] * (max_cell_lines - len(cell))) + return line_wrapped + + +# Copyright (c) 2010-2011 Gluster, Inc. <http://www.gluster.com> +# This file is part of GlusterFS. + +# GlusterFS is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published +# by the Free Software Foundation; either version 3 of the License, +# or (at your option) any later version. + +# GlusterFS is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# General Public License for more details. + +# You should have received a copy of the GNU General Public License +# along with this program. If not, see +# <http://www.gnu.org/licenses/>. + +graph_available = True + +try: + import numpy as np + import matplotlib.pyplot as plt +except ImportError: + graph_available = False + +import re +import sys + +from optparse import OptionParser + +# Global dict-of-dict holding the latency data +# latency[xlator-name][op-name] + +latencies = {} +counts = {} +totals = {} + +def collect_data (f): + """Collect latency data from the file object f and store it in + the global variable @latencies""" + + # example dump file line: + # fuse.latency.TRUNCATE=3147.000,4 + + for line in f: + m = re.search ("(\w+)\.\w+.(\w+)=(\w+\.\w+),(\w+),(\w+.\w+)", line) + if m and float(m.group(3)) != 0: + xlator = m.group(1) + op = m.group(2) + time = m.group(3) + count = m.group(4) + total = m.group(5) + + if not xlator in latencies.keys(): + latencies[xlator] = dict() + + if not xlator in counts.keys(): + counts[xlator] = dict() + + if not xlator in totals.keys(): + totals[xlator] = dict() + + latencies[xlator][op] = time + counts[xlator][op] = count + totals[xlator][op] = total + + +def calc_latency_heights (xlator_order): + heights = map (lambda x: [], xlator_order) + + N = len (xlator_order) + for i in range (N): + xl = xlator_order[i] + + k = latencies[xl].keys() + k.sort() + + if i == len (xlator_order) - 1: + # bottom-most xlator + heights[i] = [float (latencies[xl][key]) for key in k] + + else: + next_xl = xlator_order[i+1] + this_xl_time = [latencies[xl][key] for key in k] + next_xl_time = [latencies[next_xl][key] for key in k] + + heights[i] = map (lambda x, y: float (x) - float (y), + this_xl_time, next_xl_time) + return heights + +# have sufficient number of colors +colors = ["violet", "blue", "green", "yellow", "orange", "red"] + +def latency_profile (title, xlator_order, mode): + heights = calc_latency_heights (xlator_order) + + N = len (latencies[xlator_order[0]].keys()) + Nxl = len (xlator_order) + ind = np.arange (N) + width = 0.35 + + pieces = map (lambda x: [], xlator_order) + bottoms = map (lambda x: [], xlator_order) + + bottoms[Nxl-1] = map (lambda x: 0, latencies[xlator_order[0]].keys()) + + k = latencies[xlator_order[0]].keys() + k.sort() + + for i in range (Nxl-1): + xl = xlator_order[i+1] + bottoms[i] = [float(latencies[xl][key]) for key in k] + + if mode == 'text': + print "\n%sLatency profile for %s\n" % (' '*20, title) + print "Average latency (microseconds):\n" + + table = Texttable() + + table.set_cols_align(["l", "r"] + ["r"] * len(xlator_order)) + rows = [] + + header = ['OP', 'OP Average (us)'] + xlator_order + rows = [] + + for op in k: + sum = reduce (lambda x, y: x + y, [heights[xlator_order.index(xl)][k.index(op)] for xl in xlator_order], + 0) + + row = [op] + row += ["%5.2f" % sum] + + for xl in xlator_order: + op_index = k.index(op) + row += ["%5.2f" % (heights[xlator_order.index(xl)][op_index])] + + rows.append(row) + + def row_sort(r1, r2): + v1 = float(r1[1]) + v2 = float(r2[1]) + + if v1 < v2: + return -1 + elif v1 == v2: + return 0 + else: + return 1 + + rows.sort(row_sort, reverse=True) + table.add_rows([header] + rows) + print table.draw() + + elif mode == 'graph': + for i in range(Nxl): + pieces[i] = plt.bar (ind, heights[i], width, color=colors[i], + bottom=bottoms[i]) + + plt.ylabel ("Average Latency (microseconds)") + plt.title ("Latency Profile for '%s'" % title) + k = latencies[xlator_order[0]].keys() + k.sort () + plt.xticks (ind+width/2., k) + + m = round (max(map (float, latencies[xlator_order[0]].values())), -2) + plt.yticks (np.arange(0, m + m*0.1, m/10)) + plt.legend (map (lambda p: p[0], pieces), xlator_order) + + plt.show () + else: + print "Unknown mode specified!" + sys.exit(1) + + +def fop_distribution (title, xlator_order, mode): + plt.ylabel ("Percentage of calls") + plt.title ("FOP distribution for '%s'" % title) + k = counts[xlator_order[0]].keys() + k.sort () + + N = len (latencies[xlator_order[0]].keys()) + ind = np.arange(N) + width = 0.35 + + total = 0 + top_xl = xlator_order[0] + for op in k: + total += int(counts[top_xl][op]) + + heights = [] + + for op in k: + heights.append (float(counts[top_xl][op])/total * 100) + + if mode == 'text': + print "\n%sFOP distribution for %s\n" % (' '*20, title) + print "Total number of calls: %d\n" % total + + table = Texttable() + + table.set_cols_align(["l", "r", "r"]) + + rows = [] + header = ["OP", "% of Calls", "Count"] + + for op in k: + row = [op, "%5.2f" % (float(counts[top_xl][op])/total * 100), counts[top_xl][op]] + rows.append(row) + + def row_sort(r1, r2): + v1 = float(r1[1]) + v2 = float(r2[1]) + + if v1 < v2: + return -1 + elif v1 == v2: + return 0 + else: + return 1 + + rows.sort(row_sort, reverse=True) + table.add_rows([header] + rows) + print table.draw() + + elif mode == 'graph': + bars = plt.bar (ind, heights, width, color="red") + + for bar in bars: + height = bar.get_height() + plt.text (bar.get_x()+bar.get_width()/2., 1.05*height, + "%d%%" % int(height)) + + plt.xticks(ind+width/2., k) + plt.yticks(np.arange (0, 110, 10)) + + plt.show() + else: + print "mode not specified!" + sys.exit(1) + + +def calc_workload_heights (xlator_order, scaling): + workload_heights = map (lambda x: [], xlator_order) + + top_xl = xlator_order[0] + + N = len (xlator_order) + for i in range (N): + xl = xlator_order[i] + + k = totals[xl].keys() + k.sort() + + if i == len (xlator_order) - 1: + # bottom-most xlator + workload_heights[i] = [float (totals[xl][key]) / float(totals[top_xl][key]) * scaling[k.index(key)] for key in k] + + else: + next_xl = xlator_order[i+1] + this_xl_time = [float(totals[xl][key]) / float(totals[top_xl][key]) * scaling[k.index(key)] for key in k] + next_xl_time = [float(totals[next_xl][key]) / float(totals[top_xl][key]) * scaling[k.index(key)] for key in k] + + workload_heights[i] = map (lambda x, y: (float (x) - float (y)), + this_xl_time, next_xl_time) + + return workload_heights + +def workload_profile(title, xlator_order, mode): + plt.ylabel ("Percentage of Total Time") + plt.title ("Workload Profile for '%s'" % title) + k = totals[xlator_order[0]].keys() + k.sort () + + N = len(totals[xlator_order[0]].keys()) + Nxl = len(xlator_order) + ind = np.arange(N) + width = 0.35 + + total = 0 + top_xl = xlator_order[0] + for op in k: + total += float(totals[top_xl][op]) + + p_heights = [] + + for op in k: + p_heights.append (float(totals[top_xl][op])/total * 100) + + heights = calc_workload_heights (xlator_order, p_heights) + + pieces = map (lambda x: [], xlator_order) + bottoms = map (lambda x: [], xlator_order) + + bottoms[Nxl-1] = map (lambda x: 0, totals[xlator_order[0]].keys()) + + for i in range (Nxl-1): + xl = xlator_order[i+1] + k = totals[xl].keys() + k.sort() + + bottoms[i] = [float(totals[xl][key]) / float(totals[top_xl][key]) * p_heights[k.index(key)] for key in k] + + if mode == 'text': + print "\n%sWorkload profile for %s\n" % (' '*20, title) + print "Total Time: %d microseconds = %.1f seconds = %.1f minutes\n" % (total, total / 1000000.0, total / 6000000.0) + + table = Texttable() + table.set_cols_align(["l", "r"] + ["r"] * len(xlator_order)) + rows = [] + + header = ['OP', 'OP Total (%)'] + xlator_order + rows = [] + + for op in k: + sum = reduce (lambda x, y: x + y, [heights[xlator_order.index(xl)][k.index(op)] for xl in xlator_order], + 0) + row = [op] + row += ["%5.2f" % sum] + + for xl in xlator_order: + op_index = k.index(op) + row += ["%5.2f" % heights[xlator_order.index(xl)][op_index]] + + rows.append(row) + + def row_sort(r1, r2): + v1 = float(r1[1]) + v2 = float(r2[1]) + + if v1 < v2: + return -1 + elif v1 == v2: + return 0 + else: + return 1 + + rows.sort(row_sort, reverse=True) + table.add_rows([header] + rows) + print table.draw() + + elif mode == 'graph': + for i in range(Nxl): + pieces[i] = plt.bar (ind, heights[i], width, color=colors[i], + bottom=bottoms[i]) + + for key in k: + bar = pieces[Nxl-1][k.index(key)] + plt.text (bar.get_x() + bar.get_width()/2., 1.05*p_heights[k.index(key)], + "%d%%" % int(p_heights[k.index(key)])) + + plt.xticks(ind+width/2., k) + plt.yticks(np.arange (0, 110, 10)) + plt.legend (map (lambda p: p[0], pieces), xlator_order) + + plt.show() + else: + print "Unknown mode specified!" + sys.exit(1) + + +def main (): + parser = OptionParser(usage="usage: %prog [-l | -d | -w] -x <xlator order> <state dump file>") + parser.add_option("-l", "--latency", dest="latency", action="store_true", + help="Produce latency profile") + parser.add_option("-d", "--distribution", dest="distribution", action="store_true", + help="Produce distribution of FOPs") + parser.add_option("-w", "--workload", dest="workload", action="store_true", + help="Produce workload profile") + parser.add_option("-t", "--title", dest="title", help="Set the title of the graph") + parser.add_option("-x", "--xlator-order", dest="xlator_order", help="Specify the order of xlators") + parser.add_option("-m", "--mode", dest="mode", help="Output format, can be text[default] or graph") + + (options, args) = parser.parse_args() + + if len(args) != 1: + parser.error("Incorrect number of arguments") + + if (options.xlator_order): + xlator_order = options.xlator_order.split() + else: + print "xlator order must be specified" + sys.exit(1) + + collect_data(file (args[0], 'r')) + + mode = 'text' + if (options.mode): + mode = options.mode + if options.mode == 'graph' and graph_available == False: + print "matplotlib not available, falling back to text mode" + mode = 'text' + + if (options.latency): + latency_profile (options.title, xlator_order, mode) + + if (options.distribution): + fop_distribution(options.title, xlator_order, mode) + + if (options.workload): + workload_profile(options.title, xlator_order, mode) + +main () diff --git a/extras/prot_filter.py b/extras/prot_filter.py new file mode 100755 index 000000000..7dccacf15 --- /dev/null +++ b/extras/prot_filter.py @@ -0,0 +1,144 @@ +#!/usr/bin/python + +""" + Copyright (c) 2013 Red Hat, Inc. <http://www.redhat.com> + This file is part of GlusterFS. + + This file is licensed to you under your choice of the GNU Lesser + General Public License, version 3 or any later version (LGPLv3 or + later), or the GNU General Public License, version 2 (GPLv2), in all + cases as published by the Free Software Foundation. +""" + +""" + INSTRUCTIONS + Put this in /usr/lib64/glusterfs/$version/filter to have it run automatically, + or else you'll have to run it by hand every time you change the volume + configuration. Give it a list of volume names on which to enable the + protection functionality; it will deliberately ignore client volfiles for + other volumes, and all server volfiles. It *will* include internal client + volfiles such as those used for NFS or rebalance/self-heal; this is a + deliberate choice so that it will catch deletions from those sources as well. +""" + +volume_list = [ "jdtest" ] + +import copy +import string +import sys +import types + +class Translator: + def __init__ (self, name): + self.name = name + self.xl_type = "" + self.opts = {} + self.subvols = [] + self.dumped = False + def __repr__ (self): + return "<Translator %s>" % self.name + +def load (path): + # If it's a string, open it; otherwise, assume it's already a + # file-like object (most notably from urllib*). + if type(path) in types.StringTypes: + fp = file(path,"r") + else: + fp = path + all_xlators = {} + xlator = None + last_xlator = None + while True: + text = fp.readline() + if text == "": + break + text = text.split() + if not len(text): + continue + if text[0] == "volume": + if xlator: + raise RuntimeError, "nested volume definition" + xlator = Translator(text[1]) + continue + if not xlator: + raise RuntimeError, "text outside volume definition" + if text[0] == "type": + xlator.xl_type = text[1] + continue + if text[0] == "option": + xlator.opts[text[1]] = string.join(text[2:]) + continue + if text[0] == "subvolumes": + for sv in text[1:]: + xlator.subvols.append(all_xlators[sv]) + continue + if text[0] == "end-volume": + all_xlators[xlator.name] = xlator + last_xlator = xlator + xlator = None + continue + raise RuntimeError, "unrecognized keyword %s" % text[0] + if xlator: + raise RuntimeError, "unclosed volume definition" + return all_xlators, last_xlator + +def generate (graph, last, stream=sys.stdout): + for sv in last.subvols: + if not sv.dumped: + generate(graph,sv,stream) + print >> stream, "" + sv.dumped = True + print >> stream, "volume %s" % last.name + print >> stream, " type %s" % last.xl_type + for k, v in last.opts.iteritems(): + print >> stream, " option %s %s" % (k, v) + if last.subvols: + print >> stream, " subvolumes %s" % string.join( + [ sv.name for sv in last.subvols ]) + print >> stream, "end-volume" + +def push_filter (graph, old_xl, filt_type, opts={}): + new_type = "-" + filt_type.split("/")[1] + old_type = "-" + old_xl.xl_type.split("/")[1] + pos = old_xl.name.find(old_type) + if pos >= 0: + new_name = old_xl.name + old_name = new_name[:pos] + new_type + new_name[len(old_type)+pos:] + else: + new_name = old_xl.name + old_type + old_name = old_xl.name + new_type + new_xl = Translator(new_name) + new_xl.xl_type = old_xl.xl_type + new_xl.opts = old_xl.opts + new_xl.subvols = old_xl.subvols + graph[new_xl.name] = new_xl + old_xl.name = old_name + old_xl.xl_type = filt_type + old_xl.opts = opts + old_xl.subvols = [new_xl] + graph[old_xl.name] = old_xl + +if __name__ == "__main__": + path = sys.argv[1] + # Alow an override for debugging. + for extra in sys.argv[2:]: + volume_list.append(extra) + graph, last = load(path) + for v in volume_list: + if graph.has_key(v): + break + else: + print "No configured volumes found - aborting." + sys.exit(0) + for v in graph.values(): + if v.xl_type == "cluster/distribute": + push_filter(graph,v,"features/prot_dht") + elif v.xl_type == "protocol/client": + push_filter(graph,v,"features/prot_client") + # We push debug/trace so that every fop gets a real frame, because DHT + # gets confused if STACK_WIND_TAIL causes certain fops to be invoked + # from anything other than a direct child. + for v in graph.values(): + if v.xl_type == "features/prot_client": + push_filter(graph,v,"debug/trace") + generate(graph,last,stream=open(path,"w")) diff --git a/extras/quota-metadata-cleanup.sh b/extras/quota-metadata-cleanup.sh new file mode 100755 index 000000000..37ad8bc31 --- /dev/null +++ b/extras/quota-metadata-cleanup.sh @@ -0,0 +1,24 @@ +#!/bin/bash + +# This script is used to cleanup xattrs setup by quota-translator in (a) +# backend directory(ies). It takes a single or list of backend directories +# as argument(s). + +usage () +{ + echo >&2 "usage: $0 <list-of-backend-directories>" +} + +main () +{ + [ $# -lt 1 ] && usage + + INSTALL_DIR=`dirname $0` + + for i in $@; do + find $i -exec $INSTALL_DIR/quota-remove-xattr.sh '{}' \; + done + +} + +main $@ diff --git a/extras/quota-remove-xattr.sh b/extras/quota-remove-xattr.sh new file mode 100755 index 000000000..7191f9bd4 --- /dev/null +++ b/extras/quota-remove-xattr.sh @@ -0,0 +1,24 @@ +#!/bin/bash + +# This script is used to remove xattrs set by quota translator on a path. +# It is generally invoked from quota-metadata-cleanup.sh, but can +# also be used stand-alone. + +usage () +{ + echo >&2 "usage: $0 <path>" +} + +main () +{ + [ $# -ne 1 ] && usage $0 + + XATTR_KEY_VALUE_PAIRS=`getfattr -h -d -m 'trusted.glusterfs.quota' $1 2>/dev/null | sed -e '/^# file/d'` + + for i in $XATTR_KEY_VALUE_PAIRS; do + XATTR_KEY=`echo $i | sed -e 's/\([^=]*\).*/\1/g'` + setfattr -h -x $XATTR_KEY $1 + done +} + +main $@ diff --git a/extras/rebalance.py b/extras/rebalance.py new file mode 100755 index 000000000..80c614c5d --- /dev/null +++ b/extras/rebalance.py @@ -0,0 +1,299 @@ +#!/usr/bin/python + +import atexit +import copy +import optparse +import os +import pipes +import shutil +import string +import subprocess +import sys +import tempfile +import volfilter + +# It's just more convenient to have named fields. +class Brick: + def __init__ (self, path, name): + self.path = path + self.sv_name = name + self.size = 0 + self.curr_size = 0 + self.good_size = 0 + def set_size (self, size): + self.size = size + def set_range (self, rs, re): + self.r_start = rs + self.r_end = re + self.curr_size = self.r_end - self.r_start + 1 + def __repr__ (self): + value = self.path[:] + value += "(%d," % self.size + if self.curr_size: + value += "0x%x,0x%x)" % (self.r_start, self.r_end) + else: + value += "-)" + return value + +def get_bricks (host, vol): + t = pipes.Template() + t.prepend("gluster --remote-host=%s system getspec %s"%(host,vol),".-") + return t.open(None,"r") + +def generate_stanza (vf, all_xlators, cur_subvol): + sv_list = [] + for sv in cur_subvol.subvols: + generate_stanza(vf,all_xlators,sv) + sv_list.append(sv.name) + vf.write("volume %s\n"%cur_subvol.name) + vf.write(" type %s\n"%cur_subvol.type) + for kvpair in cur_subvol.opts.iteritems(): + vf.write(" option %s %s\n"%kvpair) + if sv_list: + vf.write(" subvolumes %s\n"%string.join(sv_list)) + vf.write("end-volume\n\n") + + +def mount_brick (localpath, all_xlators, dht_subvol): + + # Generate a volfile. + vf_name = localpath + ".vol" + vf = open(vf_name,"w") + generate_stanza(vf,all_xlators,dht_subvol) + vf.flush() + vf.close() + + # Create a brick directory and mount the brick there. + os.mkdir(localpath) + subprocess.call(["glusterfs","-f",vf_name,localpath]) + +# We use the command-line tools because there's no getxattr support in the +# Python standard library (which is ridiculous IMO). Adding the xattr package +# from PyPI would create a new and difficult dependency because the bits to +# satisfy it don't seem to exist in Fedora. We already expect the command-line +# tools to be there, so it's safer just to rely on them. +# +# We might have to revisit this if we get as far as actually issuing millions +# of setxattr requests. Even then, it might be better to do that part with a C +# program which has only a build-time dependency. +def get_range (brick): + t = pipes.Template() + cmd = "getfattr -e hex -n trusted.glusterfs.dht %s 2> /dev/null" + t.prepend(cmd%brick,".-") + t.append("grep ^trusted.glusterfs.dht=","--") + f = t.open(None,"r") + try: + value = f.readline().rstrip().split('=')[1][2:] + except: + print "could not get layout for %s (might be OK)" % brick + return None + v_start = int("0x"+value[16:24],16) + v_end = int("0x"+value[24:32],16) + return (v_start, v_end) + +def calc_sizes (bricks, total): + leftover = 1 << 32 + for b in bricks: + if b.size: + b.good_size = (b.size << 32) / total + leftover -= b.good_size + else: + b.good_size = 0 + if leftover: + # Add the leftover to an old brick if we can. + for b in bricks: + if b.good_size: + b.good_size += leftover + break + else: + # Fine, just add it wherever. + bricks[0].good_size += leftover + +# Normalization means sorting the bricks by r_start and (b) ensuring that there +# are no gaps. +def normalize (in_bricks): + out_bricks = [] + curr_hash = 0 + used = 0 + while curr_hash < (1<<32): + curr_best = None + for b in in_bricks: + if b.r_start == curr_hash: + used += 1 + out_bricks.append(b) + in_bricks.remove(b) + curr_hash = b.r_end + 1 + break + else: + print "gap found at 0x%08x" % curr_hash + sys.exit(1) + return out_bricks + in_bricks, used + +def get_score (bricks): + score = 0 + curr_hash = 0 + for b in bricks: + if not b.curr_size: + curr_hash += b.good_size + continue + new_start = curr_hash + curr_hash += b.good_size + new_end = curr_hash - 1 + if new_start > b.r_start: + max_start = new_start + else: + max_start = b.r_start + if new_end < b.r_end: + min_end = new_end + else: + min_end = b.r_end + if max_start <= min_end: + score += (min_end - max_start + 1) + return score + +if __name__ == "__main__": + + my_usage = "%prog [options] server volume [directory]" + parser = optparse.OptionParser(usage=my_usage) + parser.add_option("-f", "--free-space", dest="free_space", + default=False, action="store_true", + help="use free space instead of total space") + parser.add_option("-l", "--leave-mounted", dest="leave_mounted", + default=False, action="store_true", + help="leave subvolumes mounted") + parser.add_option("-v", "--verbose", dest="verbose", + default=False, action="store_true", + help="verbose output") + options, args = parser.parse_args() + + if len(args) == 3: + fix_dir = args[2] + else: + if len(args) != 2: + parser.print_help() + sys.exit(1) + fix_dir = None + hostname, volname = args[:2] + + # Make sure stuff gets cleaned up, even if there are exceptions. + orig_dir = os.getcwd() + work_dir = tempfile.mkdtemp() + bricks = [] + def cleanup_workdir (): + os.chdir(orig_dir) + if options.verbose: + print "Cleaning up %s" % work_dir + for b in bricks: + subprocess.call(["umount",b.path]) + shutil.rmtree(work_dir) + if not options.leave_mounted: + atexit.register(cleanup_workdir) + os.chdir(work_dir) + + # Mount each brick individually, so we can issue brick-specific calls. + if options.verbose: + print "Mounting subvolumes..." + index = 0 + volfile_pipe = get_bricks(hostname,volname) + all_xlators, last_xlator = volfilter.load(volfile_pipe) + for dht_vol in all_xlators.itervalues(): + if dht_vol.type == "cluster/distribute": + break + else: + print "no DHT volume found" + sys.exit(1) + for sv in dht_vol.subvols: + #print "found subvol %s" % sv.name + lpath = "%s/brick%s" % (work_dir, index) + index += 1 + mount_brick(lpath,all_xlators,sv) + bricks.append(Brick(lpath,sv.name)) + if index == 0: + print "no bricks" + sys.exit(1) + + # Collect all of the sizes. + if options.verbose: + print "Collecting information..." + total = 0 + for b in bricks: + info = os.statvfs(b.path) + # We want a standard unit even if different bricks use + # different block sizes. The size is chosen to avoid overflows + # for very large bricks with very small block sizes, but also + # accommodate filesystems which use very large block sizes to + # cheat on benchmarks. + blocksper100mb = 104857600 / info[0] + if options.free_space: + size = info[3] / blocksper100mb + else: + size = info[2] / blocksper100mb + if size <= 0: + print "brick %s has invalid size %d" % (b.path, size) + sys.exit(1) + b.set_size(size) + total += size + + # Collect all of the layout information. + for b in bricks: + hash_range = get_range(b.path) + if hash_range is not None: + rs, re = hash_range + if rs > re: + print "%s has backwards hash range" % b.path + sys.exit(1) + b.set_range(hash_range[0],hash_range[1]) + + if options.verbose: + print "Calculating new layouts..." + calc_sizes(bricks,total) + bricks, used = normalize(bricks) + + # We can't afford O(n!) here, but O(n^2) should be OK and the result + # should be almost as good. + while used < len(bricks): + best_place = used + best_score = get_score(bricks) + for i in xrange(used): + new_bricks = bricks[:] + del new_bricks[used] + new_bricks.insert(i,bricks[used]) + new_score = get_score(new_bricks) + if new_score > best_score: + best_place = i + best_score = new_score + if best_place != used: + nb = bricks[used] + del bricks[used] + bricks.insert(best_place,nb) + used += 1 + + # Finalize whatever we decided on. + curr_hash = 0 + for b in bricks: + b.r_start = curr_hash + curr_hash += b.good_size + b.r_end = curr_hash - 1 + + print "Here are the xattr values for your size-weighted layout:" + for b in bricks: + print " %s: 0x0000000200000000%08x%08x" % ( + b.sv_name, b.r_start, b.r_end) + + if fix_dir: + if options.verbose: + print "Fixing layout for %s" % fix_dir + for b in bricks: + value = "0x0000000200000000%08x%08x" % ( + b.r_start, b.r_end) + path = "%s/%s" % (b.path, fix_dir) + cmd = "setfattr -n trusted.glusterfs.dht -v %s %s" % ( + value, path) + print cmd + + if options.leave_mounted: + print "The following subvolumes are still mounted:" + for b in bricks: + print "%s on %s" % (b.sv_name, b.path) + print "Don't forget to clean up when you're done." + diff --git a/extras/rpc-coverage.sh b/extras/rpc-coverage.sh new file mode 100755 index 000000000..9148a73e2 --- /dev/null +++ b/extras/rpc-coverage.sh @@ -0,0 +1,480 @@ +#!/bin/bash + +# This script can be used to provoke 35 fops (if afr is used), +# 28 fops (if afr is not used) (-fstat,-readdirp, and lk,xattrop calls) +# Pending are 7 procedures. +# getspec, fsyncdir, access, fentrylk, fsetxattr, fgetxattr, rchecksum +# TODO: add commands which can generate fops for missing fops + +## Script tests below File Operations over RPC (when afr is used) + +# CREATE +# ENTRYLK +# FINODELK +# FLUSH +# FSTAT +# FSYNC +# FTRUNCATE +# FXATTROP +# GETXATTR +# INODELK +# LINK +# LK +# LOOKUP +# MKDIR +# MKNOD +# OPEN +# OPENDIR +# READ +# READDIR +# READDIRP +# READLINK +# RELEASE +# RELEASEDIR +# REMOVEXATTR +# RENAME +# RMDIR +# SETATTR +# SETXATTR +# STAT +# STATFS +# SYMLINK +# TRUNCATE +# UNLINK +# WRITE +# XATTROP + +#set -e; +set -o pipefail; + +function fail() { + echo "$*: failed."; + exit 1; +} + +function test_mkdir() +{ + mkdir -p $PFX/dir; + test $(stat -c '%F' $PFX/dir) == "directory" || fail "mkdir" +} + + +function test_create() +{ + : > $PFX/dir/file; + + test "$(stat -c '%F' $PFX/dir/file)" == "regular empty file" || fail "create" +} + + +function test_statfs() +{ + local size; + + size=$(stat -f -c '%s' $PFX/dir/file); + test "x$size" != "x0" || fail "statfs" +} + + +function test_open() +{ + exec 4<$PFX/dir/file || fail "open" + exec 4>&- || fail "open" +} + + +function test_write() +{ + dd if=/dev/zero of=$PFX/dir/file bs=65536 count=16 2>/dev/null; + test $(stat -c '%s' $PFX/dir/file) == 1048576 || fail "open" +} + + +function test_read() +{ + local count; + + count=$(dd if=$PFX/dir/file bs=64k count=16 2>/dev/null | wc -c); + test $count == 1048576 || fail "read" +} + + +function test_truncate() +{ + truncate -s 512 $PFX/dir/file; + test $(stat -c '%s' $PFX/dir/file) == 512 || fail "truncate" + + truncate -s 0 $PFX/dir/file; + test $(stat -c '%s' $PFX/dir/file) == 0 || fail "truncate" +} + + +function test_fstat() +{ + local msg; + + export PFX; + msg=$(sh -c 'tail -f $PFX/dir/file --pid=$$ & sleep 1 && echo hooha > $PFX/dir/file && sleep 1'); + test "x$msg" == "xhooha" || fail "fstat" +} + + +function test_mknod() +{ + mknod -m 0666 $PFX/dir/block b 13 42; + test "$(stat -c '%F %a %t %T' $PFX/dir/block)" == "block special file 666 d 2a" \ + || fail "mknod for block device" + + mknod -m 0666 $PFX/dir/char c 13 42; + test "$(stat -c '%F %a %t %T' $PFX/dir/char)" == "character special file 666 d 2a" \ + || fail "mknod for character device" + + mknod -m 0666 $PFX/dir/fifo p; + test "$(stat -c '%F %a' $PFX/dir/fifo)" == "fifo 666" || \ + fail "mknod for fifo" +} + + +function test_symlink() +{ + local msg; + + ln -s $PFX/dir/file $PFX/dir/symlink; + test "$(stat -c '%F' $PFX/dir/symlink)" == "symbolic link" || fail "Creation of symlink" + + msg=$(cat $PFX/dir/symlink); + test "x$msg" == "xhooha" || fail "Content match for symlink" +} + + +function test_hardlink() +{ + local ino1; + local ino2; + local nlink1; + local nlink2; + local msg; + + ln $PFX/dir/file $PFX/dir/hardlink; + + ino1=$(stat -c '%i' $PFX/dir/file); + nlink1=$(stat -c '%h' $PFX/dir/file); + ino2=$(stat -c '%i' $PFX/dir/hardlink); + nlink2=$(stat -c '%h' $PFX/dir/hardlink); + + test $ino1 == $ino2 || fail "Inode comparison for hardlink" + test $nlink1 == 2 || fail "Link count for hardlink" + test $nlink2 == 2 || fail "Link count for hardlink" + + msg=$(cat $PFX/dir/hardlink); + + test "x$msg" == "xhooha" || fail "Content match for hardlinks" +} + + +function test_rename() +{ + local ino1; + local ino2; + local ino3; + local msg; + + #### file + + ino1=$(stat -c '%i' $PFX/dir/file); + + mv $PFX/dir/file $PFX/dir/file2 || fail "mv" + msg=$(cat $PFX/dir/file2); + test "x$msg" == "xhooha" || fail "File contents comparison after mv" + + ino2=$(stat -c '%i' $PFX/dir/file2); + test $ino1 == $ino2 || fail "Inode comparison after mv" + + mv $PFX/dir/file2 $PFX/dir/file; + msg=$(cat $PFX/dir/file); + test "x$msg" == "xhooha" || fail "File contents comparison after mv" + + ino3=$(stat -c '%i' $PFX/dir/file); + test $ino1 == $ino3 || fail "Inode comparison after mv" + + #### dir + + ino1=$(stat -c '%i' $PFX/dir); + + mv $PFX/dir $PFX/dir2 || fail "Directory mv" + ino2=$(stat -c '%i' $PFX/dir2); + test $ino1 == $ino2 || fail "Inode comparison after directory mv" + + mv $PFX/dir2 $PFX/dir || fail "Directory mv" + ino3=$(stat -c '%i' $PFX/dir); + test $ino1 == $ino3 || fail "Inode comparison after directory mv" +} + + +function test_chmod() +{ + local mode0; + local mode1; + local mode2; + + + #### file + + mode0=$(stat -c '%a' $PFX/dir/file); + chmod 0753 $PFX/dir/file || fail "chmod" + + mode1=$(stat -c '%a' $PFX/dir/file); + test 0$mode1 == 0753 || fail "Mode comparison after chmod" + + chmod 0$mode0 $PFX/dir/file || fail "chmod" + mode2=$(stat -c '%a' $PFX/dir/file); + test 0$mode2 == 0$mode0 || fail "Mode comparison after chmod" + + #### dir + + mode0=$(stat -c '%a' $PFX/dir); + chmod 0753 $PFX/dir || fail "chmod" + + mode1=$(stat -c '%a' $PFX/dir); + test 0$mode1 == 0753 || fail "Mode comparison after chmod" + + chmod 0$mode0 $PFX/dir || fail "chmod" + mode2=$(stat -c '%a' $PFX/dir); + test 0$mode2 == 0$mode0 || fail "Mode comparison after chmod" +} + + +function test_chown() +{ + local user1; + local user2; + local group1; + local group2; + + #### file + + user1=$(stat -c '%u' $PFX/dir/file); + group1=$(stat -c '%g' $PFX/dir/file); + + chown 13:42 $PFX/dir/file || fail "chown" + + user2=$(stat -c '%u' $PFX/dir/file); + group2=$(stat -c '%g' $PFX/dir/file); + + test $user2 == 13 || fail "User comparison after chown" + test $group2 == 42 || fail "Group comparison after chown" + + chown $user1:$group1 $PFX/dir/file || fail "chown" + + user2=$(stat -c '%u' $PFX/dir/file); + group2=$(stat -c '%g' $PFX/dir/file); + + test $user2 == $user1 || fail "User comparison after chown" + test $group2 == $group1 || fail "Group comparison after chown" + + #### dir + + user1=$(stat -c '%u' $PFX/dir); + group1=$(stat -c '%g' $PFX/dir); + + chown 13:42 $PFX/dir || fail "chown" + + user2=$(stat -c '%u' $PFX/dir); + group2=$(stat -c '%g' $PFX/dir); + + test $user2 == 13 || fail "User comparison after chown" + test $group2 == 42 || fail "Group comparison after chown" + + chown $user1:$group1 $PFX/dir || fail "chown" + + user2=$(stat -c '%u' $PFX/dir); + group2=$(stat -c '%g' $PFX/dir); + + test $user2 == $user1 || fail "User comparison after chown" + test $group2 == $group1 || fail "Group comparison after chown" +} + + +function test_utimes() +{ + local acc0; + local acc1; + local acc2; + local mod0; + local mod1; + local mod2; + + #### file + + acc0=$(stat -c '%X' $PFX/dir/file); + mod0=$(stat -c '%Y' $PFX/dir/file); + + sleep 1; + touch -a $PFX/dir/file || fail "atime change on file" + + acc1=$(stat -c '%X' $PFX/dir/file); + mod1=$(stat -c '%Y' $PFX/dir/file); + + sleep 1; + touch -m $PFX/dir/file || fail "mtime change on file" + + acc2=$(stat -c '%X' $PFX/dir/file); + mod2=$(stat -c '%Y' $PFX/dir/file); + + test $acc0 != $acc1 || fail "atime mismatch comparison on file" + test $acc1 == $acc2 || fail "atime match comparison on file" + test $mod0 == $mod1 || fail "mtime match comparison on file" + test $mod1 != $mod2 || fail "mtime mismatch comparison on file" + + #### dir + + acc0=$(stat -c '%X' $PFX/dir); + mod0=$(stat -c '%Y' $PFX/dir); + + sleep 1; + touch -a $PFX/dir || fail "atime change on directory" + + acc1=$(stat -c '%X' $PFX/dir); + mod1=$(stat -c '%Y' $PFX/dir); + + sleep 1; + touch -m $PFX/dir || fail "mtime change on directory" + + acc2=$(stat -c '%X' $PFX/dir); + mod2=$(stat -c '%Y' $PFX/dir); + + test $acc0 != $acc1 || fail "atime mismatch comparison on directory" + test $acc1 == $acc2 || fail "atime match comparison on directory" + test $mod0 == $mod1 || fail "mtime match comparison on directory" + test $mod1 != $mod2 || fail "mtime mismatch comparison on directory" +} + + +function test_locks() +{ + exec 200>$PFX/dir/lockfile || fail "exec" + + ## exclusive locks test + flock -e 200 || fail "flock -e" + ! flock -n -e $PFX/dir/lockfile -c true || fail "! flock -n -e" + ! flock -n -s $PFX/dir/lockfile -c true || fail "! flock -n -s" + flock -u 200 || fail "flock -u" + + ## shared locks test + flock -s 200 || fail "flock -s" + ! flock -n -e $PFX/dir/lockfile -c true || fail "! flock -n -e" + flock -n -s $PFX/dir/lockfile -c true || fail "! flock -n -s" + flock -u 200 || fail "flock -u" + + exec 200>&- || fail "exec" + +} + + +function test_readdir() +{ + /bin/ls $PFX/dir >/dev/null || fail "ls" +} + + +function test_setxattr() +{ + setfattr -n trusted.testing -v c00k33 $PFX/dir/file || fail "setfattr" +} + + +function test_listxattr() +{ + getfattr -m trusted $PFX/dir/file 2>/dev/null | grep -q trusted.testing || fail "getfattr" +} + + +function test_getxattr() +{ + getfattr -n trusted.testing $PFX/dir/file 2>/dev/null | grep -q c00k33 || fail "getfattr" +} + + +function test_removexattr() +{ + setfattr -x trusted.testing $PFX/dir/file || fail "setfattr remove" + getfattr -n trusted.testing $PFXf/dir/file 2>&1 | grep -q "No such attribute" +} + + +function test_unlink() +{ + rm $PFX/dir/file || fail "rm" +} + + +function test_rmdir() +{ + rm -rf $PFX || fail "rm -rf" +} + + +function run_tests() +{ + test_mkdir; + test_create; + test_statfs; + test_open; + test_write; + test_read; + test_truncate; + test_fstat; + test_mknod; + test_hardlink; + test_symlink; + test_rename; + test_chmod; + test_chown; + test_utimes; + test_locks; + test_readdir; + test_setxattr; + test_listxattr; + test_getxattr; + test_removexattr; + test_unlink; + test_rmdir; +} + + +function _init() +{ + DIR=$(pwd); +} + + +function parse_cmdline() +{ + if [ "x$1" == "x" ] ; then + echo "Usage: $0 /path/mount" + exit 1 + fi + + DIR=$1; + + if [ ! -d "$DIR" ] ; then + echo "$DIR: not a directory" + exit 1 + fi + + PFX="$DIR/coverage"; + rm -rvf $PFX; +} + + +function main() +{ + parse_cmdline "$@"; + + run_tests; + + exit 0; +} + + +_init && main "$@"; diff --git a/extras/specgen.scm b/extras/specgen.scm index 32aff6480..0edbb6f62 100755 --- a/extras/specgen.scm +++ b/extras/specgen.scm @@ -1,7 +1,7 @@ #!/usr/bin/guile -s !# -;;; Copyright (C) 2007-2009 Z RESEARCH Inc. <http://www.zresearch.com> +;;; Copyright (c) 2007-2011 Gluster Inc. <http://www.gluster.com> ;;; ;;; This program is free software; you can redistribute it and/or modify ;;; it under the terms of the GNU General Public License as published by diff --git a/extras/stripe-merge.c b/extras/stripe-merge.c index 3f8e4b124..74bd47e30 100644 --- a/extras/stripe-merge.c +++ b/extras/stripe-merge.c @@ -1,48 +1,494 @@ +/* + Copyright (c) 2008-2012 Red Hat, Inc. <http://www.redhat.com> + This file is part of GlusterFS. + + This file is licensed to you under your choice of the GNU Lesser + General Public License, version 3 or any later version (LGPLv3 or + later), or the GNU General Public License, version 2 (GPLv2), in all + cases as published by the Free Software Foundation. +*/ + +/* + * stripe-merge.c + * + * This program recovers an original file based on the striped files stored on + * the individual bricks of a striped volume. The file format and stripe + * geometry is validated through the extended attributes stored in the file. + * + * TODO: Support optional xattr recovery (i.e., user xattrs). Perhaps provide a + * command-line flag to toggle this behavior. + */ + #include <stdio.h> -#include <unistd.h> -#include <fcntl.h> #include <sys/types.h> #include <sys/stat.h> +#include <fcntl.h> +#include <unistd.h> +#include <stdlib.h> +#include <stdint.h> +#include <errno.h> +#include <string.h> +#include <attr/xattr.h> +#include <fnmatch.h> -int -main (int argc, char *argv[]) +#define ATTRNAME_STRIPE_INDEX "trusted.*.stripe-index" +#define ATTRNAME_STRIPE_COUNT "trusted.*.stripe-count" +#define ATTRNAME_STRIPE_SIZE "trusted.*.stripe-size" +#define ATTRNAME_STRIPE_COALESCE "trusted.*.stripe-coalesce" + +#define INVALID_FD -1 +#define INVALID_MODE UINT32_MAX + +struct file_stripe_info { + int stripe_count; + int stripe_size; + int coalesce; + mode_t mode; + int fd[0]; +}; + +static int close_files(struct file_stripe_info *); + +static struct +file_stripe_info *alloc_file_stripe_info(int count) { - int fds[argc-1]; - char buf[argc-1][4096]; int i; - int max_ret, ret; + struct file_stripe_info *finfo; - if (argc < 2) { - printf ("Usage: %s file1 file2 ... >file\n", argv[0]); - return 1; + finfo = calloc(1, sizeof(struct file_stripe_info) + + (sizeof(int) * count)); + if (!finfo) + return NULL; + + for (i = 0; i < count; i++) + finfo->fd[i] = INVALID_FD; + + finfo->mode = INVALID_MODE; + finfo->coalesce = INVALID_FD; + + return finfo; +} + +/* + * Search for an attribute matching the provided pattern. Return a count for + * the total number of matching entries (including 0). Allocate a buffer for + * the first matching entry found. + */ +static int +get_stripe_attr_name(const char *path, const char *pattern, char **attrname) +{ + char attrbuf[4096]; + char *ptr, *match = NULL; + int len, r, match_count = 0; + + if (!path || !pattern || !attrname) + return -1; + + len = listxattr(path, attrbuf, sizeof(attrbuf)); + if (len < 0) + return len; + + ptr = attrbuf; + while (ptr) { + r = fnmatch(pattern, ptr, 0); + if (!r) { + if (!match) + match = ptr; + match_count++; + } else if (r != FNM_NOMATCH) { + return -1; + } + + len -= strlen(ptr) + 1; + if (len > 0) + ptr += strlen(ptr) + 1; + else + ptr = NULL; } - for (i=0; i<argc-1; i++) { - fds[i] = open (argv[i+1], O_RDONLY); - if (fds[i] == -1) { - perror (argv[i+1]); - return 1; + if (match) + *attrname = strdup(match); + + return match_count; +} + +/* + * Get the integer representation of a named attribute. + */ +static int +get_stripe_attr_val(const char *path, const char *attr, int *val) +{ + char attrbuf[4096]; + int len; + + if (!path || !attr || !val) + return -1; + + len = getxattr(path, attr, attrbuf, sizeof(attrbuf)); + if (len < 0) + return len; + + *val = atoi(attrbuf); + + return 0; +} + +/* + * Get an attribute name/value (assumed to be an integer) pair based on a + * specified search pattern. A buffer is allocated for the exact attr name + * returned. Optionally, skip the pattern search if a buffer is provided + * (which should contain an attribute name). + * + * Returns the attribute count or -1 on error. The value parameter is set only + * when a single attribute is found. + */ +static int +get_attr(const char *path, const char *pattern, char **buf, int *val) +{ + int count = 1; + + if (!buf) + return -1; + + if (!*buf) { + count = get_stripe_attr_name(path, pattern, buf); + if (count > 1) { + /* pattern isn't good enough */ + fprintf(stderr, "ERROR: duplicate attributes found " + "matching pattern: %s\n", pattern); + free(*buf); + *buf = NULL; + return count; + } else if (count < 1) { + return count; } } - max_ret = 0; + if (get_stripe_attr_val(path, *buf, val) < 0) + return -1; + + return count; +} + +/* + * validate_and_open_files() + * + * Open the provided source files and validate the extended attributes. Verify + * that the geometric attributes are consistent across all of the files and + * print a warning if any files are missing. We proceed without error in the + * latter case to support partial recovery. + */ +static struct +file_stripe_info *validate_and_open_files(char *paths[], int count) +{ + int i, val, tmp; + struct stat sbuf; + char *stripe_count_attr = NULL; + char *stripe_size_attr = NULL; + char *stripe_index_attr = NULL; + char *stripe_coalesce_attr = NULL; + struct file_stripe_info *finfo = NULL; + + for (i = 0; i < count; i++) { + if (!paths[i]) + goto err; + + /* + * Check the stripe count first so we can allocate the info + * struct with the appropriate number of fds. + */ + if (get_attr(paths[i], ATTRNAME_STRIPE_COUNT, + &stripe_count_attr, &val) != 1) { + fprintf(stderr, "ERROR: %s: attribute: '%s'\n", + paths[i], ATTRNAME_STRIPE_COUNT); + goto err; + } + if (!finfo) { + finfo = alloc_file_stripe_info(val); + if (!finfo) + goto err; + + if (val != count) + fprintf(stderr, "WARNING: %s: stripe-count " + "(%d) != file count (%d). Result may " + "be incomplete.\n", paths[i], val, + count); + + finfo->stripe_count = val; + } else if (val != finfo->stripe_count) { + fprintf(stderr, "ERROR %s: invalid stripe count: %d " + "(expected %d)\n", paths[i], val, + finfo->stripe_count); + goto err; + } + + /* + * Get and validate the chunk size. + */ + if (get_attr(paths[i], ATTRNAME_STRIPE_SIZE, &stripe_size_attr, + &val) != 1) { + fprintf(stderr, "ERROR: %s: attribute: '%s'\n", + paths[i], ATTRNAME_STRIPE_SIZE); + goto err; + } + + if (!finfo->stripe_size) { + finfo->stripe_size = val; + } else if (val != finfo->stripe_size) { + fprintf(stderr, "ERROR: %s: invalid stripe size: %d " + "(expected %d)\n", paths[i], val, + finfo->stripe_size); + goto err; + } + + /* + * stripe-coalesce is a backward compatible attribute. If the + * attribute does not exist, assume a value of zero for the + * traditional stripe format. + */ + tmp = get_attr(paths[i], ATTRNAME_STRIPE_COALESCE, + &stripe_coalesce_attr, &val); + if (!tmp) { + val = 0; + } else if (tmp != 1) { + fprintf(stderr, "ERROR: %s: attribute: '%s'\n", + paths[i], ATTRNAME_STRIPE_COALESCE); + goto err; + } + + if (finfo->coalesce == INVALID_FD) { + finfo->coalesce = val; + } else if (val != finfo->coalesce) { + fprintf(stderr, "ERROR: %s: invalid coalesce flag\n", + paths[i]); + goto err; + } + + /* + * Get/validate the stripe index and open the file in the + * appropriate fd slot. + */ + if (get_attr(paths[i], ATTRNAME_STRIPE_INDEX, + &stripe_index_attr, &val) != 1) { + fprintf(stderr, "ERROR: %s: attribute: '%s'\n", + paths[i], ATTRNAME_STRIPE_INDEX); + goto err; + } + if (finfo->fd[val] != INVALID_FD) { + fprintf(stderr, "ERROR: %s: duplicate stripe index: " + "%d\n", paths[i], val); + goto err; + } + + finfo->fd[val] = open(paths[i], O_RDONLY); + if (finfo->fd[val] < 0) + goto err; + + /* + * Get the creation mode for the file. + */ + if (fstat(finfo->fd[val], &sbuf) < 0) + goto err; + if (finfo->mode == INVALID_MODE) { + finfo->mode = sbuf.st_mode; + } else if (sbuf.st_mode != finfo->mode) { + fprintf(stderr, "ERROR: %s: invalid mode\n", paths[i]); + goto err; + } + } + + free(stripe_count_attr); + free(stripe_size_attr); + free(stripe_index_attr); + free(stripe_coalesce_attr); + + return finfo; +err: + + free(stripe_count_attr); + free(stripe_size_attr); + free(stripe_index_attr); + free(stripe_coalesce_attr); + + if (finfo) { + close_files(finfo); + free(finfo); + } + + return NULL; +} + +static int +close_files(struct file_stripe_info *finfo) +{ + int i, ret; + + if (!finfo) + return -1; + + for (i = 0; i < finfo->stripe_count; i++) { + if (finfo->fd[i] == INVALID_FD) + continue; + + ret = close(finfo->fd[i]); + if (ret < 0) + return ret; + } + + return ret; +} + +/* + * Generate the original file using files striped in the coalesced format. + * Data in the striped files is stored at a coalesced offset based on the + * stripe number. + * + * Walk through the finfo fds (which are already ordered) and and iteratively + * copy stripe_size bytes from the source files to the target file. If a source + * file is missing, seek past the associated stripe_size bytes in the target + * file. + */ +static int +generate_file_coalesce(int target, struct file_stripe_info *finfo) +{ + char *buf; + int ret = 0; + int r, w, i; + + buf = malloc(finfo->stripe_size); + if (!buf) + return -1; + + i = 0; + while (1) { + if (finfo->fd[i] == INVALID_FD) { + if (lseek(target, finfo->stripe_size, SEEK_CUR) < 0) + break; + + i = (i + 1) % finfo->stripe_count; + continue; + } + + r = read(finfo->fd[i], buf, finfo->stripe_size); + if (r < 0) { + ret = r; + break; + } + if (!r) + break; + + w = write(target, buf, r); + if (w < 0) { + ret = w; + break; + } + + i = (i + 1) % finfo->stripe_count; + } + + free(buf); + return ret; +} + +/* + * Generate the original file using files striped with the traditional stripe + * format. Data in the striped files is stored at the equivalent offset from + * the source file. + */ +static int +generate_file_traditional(int target, struct file_stripe_info *finfo) +{ + int i, j, max_ret, ret; + char buf[finfo->stripe_count][4096]; + do { char newbuf[4096] = {0, }; - int j; max_ret = 0; - for (i=0; i<argc-1; i++) { - memset (buf[i], 0, 4096); - ret = read (fds[i], buf[i], 4096); + for (i = 0; i < finfo->stripe_count; i++) { + memset(buf[i], 0, 4096); + ret = read(finfo->fd[i], buf[i], 4096); if (ret > max_ret) max_ret = ret; } - for (i=0; i<max_ret;i++) - for (j=0; j<argc-1; j++) + for (i = 0; i < max_ret; i++) + for (j = 0; j < finfo->stripe_count; j++) newbuf[i] |= buf[j][i]; - write (1, newbuf, max_ret); + write(target, newbuf, max_ret); } while (max_ret); return 0; } +static int +generate_file(int target, struct file_stripe_info *finfo) +{ + if (finfo->coalesce) + return generate_file_coalesce(target, finfo); + + return generate_file_traditional(target, finfo); +} + +static void +usage(char *name) +{ + fprintf(stderr, "Usage: %s [-o <outputfile>] <inputfile1> " + "<inputfile2> ...\n", name); +} + +int +main(int argc, char *argv[]) +{ + int file_count, opt; + char *opath = NULL; + int targetfd; + struct file_stripe_info *finfo; + + while ((opt = getopt(argc, argv, "o:")) != -1) { + switch (opt) { + case 'o': + opath = optarg; + break; + default: + usage(argv[0]); + return -1; + } + } + + file_count = argc - optind; + + if (!opath || !file_count) { + usage(argv[0]); + return -1; + } + + finfo = validate_and_open_files(&argv[optind], file_count); + if (!finfo) + goto err; + + targetfd = open(opath, O_RDWR|O_CREAT, finfo->mode); + if (targetfd < 0) + goto err; + + if (generate_file(targetfd, finfo) < 0) + goto err; + + if (fsync(targetfd) < 0) + fprintf(stderr, "ERROR: %s\n", strerror(errno)); + if (close(targetfd) < 0) + fprintf(stderr, "ERROR: %s\n", strerror(errno)); + + close_files(finfo); + free(finfo); + + return 0; + +err: + if (finfo) { + close_files(finfo); + free(finfo); + } + + return -1; +} + diff --git a/extras/systemd/Makefile.am b/extras/systemd/Makefile.am new file mode 100644 index 000000000..3fc656b82 --- /dev/null +++ b/extras/systemd/Makefile.am @@ -0,0 +1,11 @@ + +CLEANFILES = + +SYSTEMD_DIR = @systemddir@ + +install-exec-local: + @if [ -d $(SYSTEMD_DIR) ]; then \ + $(mkdir_p) $(DESTDIR)$(SYSTEMD_DIR); \ + $(INSTALL_PROGRAM) glusterd.service $(DESTDIR)$(SYSTEMD_DIR)/; \ + fi + diff --git a/extras/systemd/glusterd.service.in b/extras/systemd/glusterd.service.in new file mode 100644 index 000000000..fc8d8c9a2 --- /dev/null +++ b/extras/systemd/glusterd.service.in @@ -0,0 +1,14 @@ +[Unit] +Description=GlusterFS, a clustered file-system server +After=network.target rpcbind.service +Before=network-online.target + +[Service] +Type=forking +PIDFile=/run/glusterd.pid +LimitNOFILE=65536 +ExecStart=@prefix@/sbin/glusterd -p /run/glusterd.pid +KillMode=process + +[Install] +WantedBy=multi-user.target diff --git a/extras/test/bug-920583.t b/extras/test/bug-920583.t new file mode 100755 index 000000000..eedbb800a --- /dev/null +++ b/extras/test/bug-920583.t @@ -0,0 +1,50 @@ +#!/bin/bash + +##Copy this file to tests/bugs before running run.sh (cp extras/test/bug-920583.t tests/bugs/) + +. $(dirname $0)/../include.rc +. $(dirname $0)/../volume.rc + +cleanup; +logdir=`gluster --print-logdir` + +## Start and create a volume +TEST glusterd; +TEST pidof glusterd; + +TEST $CLI volume create $V0 replica 2 stripe 2 $H0:$B0/${V0}{1,2,3,4,5,6,7,8}; + +## Verify volume is is created +EXPECT "$V0" volinfo_field $V0 'Volume Name'; +EXPECT 'Created' volinfo_field $V0 'Status'; + +## Start volume and verify +TEST $CLI volume start $V0; +EXPECT 'Started' volinfo_field $V0 'Status'; + +function log-file-name() +{ + logfilename=$M0".log" + echo ${logfilename:1} | tr / - +} + +log_file=$logdir"/"`log-file-name` + +lookup_unhashed_count=`grep "adding option 'lookup-unhashed'" $log_file | wc -l` +no_child_down_count=`grep "adding option 'assert-no-child-down'" $log_file | wc -l` +mount -t glusterfs $H0:/$V0 $M0 -o "xlator-option=*dht.assert-no-child-down=yes,xlator-option=*dht.lookup-unhashed=yes" +touch $M0/file1; + +new_lookup_unhashed_count=`grep "adding option 'lookup-unhashed'" $log_file | wc -l` +new_no_child_down_count=`grep "adding option 'assert-no-child-down'" $log_file | wc -l` +EXPECT "1" expr $new_lookup_unhashed_count - $lookup_unhashed_count +EXPECT "1" expr $new_no_child_down_count - $no_child_down_count + +## Finish up +TEST $CLI volume stop $V0; +EXPECT 'Stopped' volinfo_field $V0 'Status'; + +TEST $CLI volume delete $V0; +TEST ! $CLI volume info $V0; + +cleanup; diff --git a/extras/test/gluster_commands.sh b/extras/test/gluster_commands.sh new file mode 100755 index 000000000..cb2a55fd5 --- /dev/null +++ b/extras/test/gluster_commands.sh @@ -0,0 +1,265 @@ +#!/bin/bash + + +# Copyright (c) 2006-2012 Red Hat, Inc. <http://www.redhat.com> +# This file is part of GlusterFS. +# +# This file is licensed to you under your choice of the GNU Lesser +# General Public License, version 3 or any later version (LGPLv3 or +# later), or the GNU General Public License, version 2 (GPLv2), in all +# cases as published by the Free Software Foundation. + + +# This script tests the basics gluster cli commands. + +echo "Starting glusterd" +glusterd +if [ $? -ne 0 ]; then + echo "Could not start glusterd.Exiting" + exit; +else + echo "glusterd started" +fi + +if [ ! -d "/exports" ]; then + mkdir /exports; + mkdir /exports/exp{1..10}; +else + mkdir /exports/exp{1..10}; +fi + +if [ ! -d "/mnt/client" ]; then + mkdir /mnt/client -p; +fi + + +set -e #exit at the first error that happens + +# create distribute volume and try start, mount, add-brick, replace-brick, remove-brick, stop, unmount, delete + +echo "Creating distribute volume........" +gluster volume create vol `hostname`:/exports/exp1 +gluster volume info + +echo "Starting distribute volume........" +gluster volume start vol +gluster volume info +sleep 1 +mount -t glusterfs `hostname`:vol /mnt/client +sleep 1 +df -h + +echo "adding-brick......." +gluster volume add-brick vol `hostname`:/exports/exp2 +gluster volume info +sleep 1 +umount /mnt/client +mount -t glusterfs `hostname`:vol /mnt/client +df -h +sleep 1 + +echo "replacing brick......" +gluster volume replace-brick vol `hostname`:/exports/exp1 `hostname`:/exports/exp3 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick vol `hostname`:/exports/exp1 `hostname`:/exports/exp3 status +gluster volume replace-brick vol `hostname`:/exports/exp1 `hostname`:/exports/exp3 pause +gluster volume replace-brick vol `hostname`:/exports/exp1 `hostname`:/exports/exp3 status +gluster volume replace-brick vol `hostname`:/exports/exp1 `hostname`:/exports/exp3 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick vol `hostname`:/exports/exp1 `hostname`:/exports/exp3 status +gluster volume replace-brick vol `hostname`:/exports/exp1 `hostname`:/exports/exp3 commit + + +echo "replcing brick for abort operation" +gluster volume replace-brick vol `hostname`:/exports/exp3 `hostname`:/exports/exp1 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick vol `hostname`:/exports/exp3 `hostname`:/exports/exp1 status +gluster volume replace-brick vol `hostname`:/exports/exp3 `hostname`:/exports/exp1 pause +gluster volume replace-brick vol `hostname`:/exports/exp3 `hostname`:/exports/exp1 status +gluster volume replace-brick vol `hostname`:/exports/exp3 `hostname`:/exports/exp1 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick vol `hostname`:/exports/exp3 `hostname`:/exports/exp1 status +gluster volume replace-brick vol `hostname`:/exports/exp3 `hostname`:/exports/exp1 abort + + +gluster volume info +sleep 1 +df -h +sleep 1 + +echo "removing brick......." +gluster --mode=script volume remove-brick vol `hostname`:/exports/exp2 +gluster volume info +sleep 1 +df -h +sleep 1 + +echo "stopping distribute volume......" +gluster --mode=script volume stop vol +gluster volume info +sleep 1 +umount /mnt/client +df -h + +echo "deleting distribute volume......" +gluster --mode=script volume delete vol +gluster volume info +sleep 1 + +# create replicate volume and try start, mount, add-brick, replace-brick, remove-brick, stop, unmount, delete +echo "creating replicate volume......" +gluster volume create mirror replica 2 `hostname`:/exports/exp1 `hostname`:/exports/exp2 +gluster volume info +sleep 1 + +echo "starting replicate volume......" +gluster volume start mirror +gluster volume info +sleep 1 +mount -t glusterfs `hostname`:mirror /mnt/client +sleep 1 +df -h +sleep 1 + +echo "adding-brick......." +gluster volume add-brick mirror `hostname`:/exports/exp3 `hostname`:/exports/exp4 +gluster volume info +sleep 1 +df -h +sleep 1 + +echo "replacing-brick....." +gluster volume replace-brick mirror `hostname`:/exports/exp1 `hostname`:/exports/exp5 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick mirror `hostname`:/exports/exp1 `hostname`:/exports/exp5 status +gluster volume replace-brick mirror `hostname`:/exports/exp1 `hostname`:/exports/exp5 pause +gluster volume replace-brick mirror `hostname`:/exports/exp1 `hostname`:/exports/exp5 status +gluster volume replace-brick mirror `hostname`:/exports/exp1 `hostname`:/exports/exp5 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick mirror `hostname`:/exports/exp1 `hostname`:/exports/exp5 status +gluster volume replace-brick mirror `hostname`:/exports/exp1 `hostname`:/exports/exp5 commit +gluster volume info +sleep 1 +df -h +sleep 1 + +echo "replacing brick for abort operation" +gluster volume replace-brick mirror `hostname`:/exports/exp5 `hostname`:/exports/exp1 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick mirror `hostname`:/exports/exp5 `hostname`:/exports/exp1 status +gluster volume replace-brick mirror `hostname`:/exports/exp5 `hostname`:/exports/exp1 pause +gluster volume replace-brick mirror `hostname`:/exports/exp5 `hostname`:/exports/exp1 status +gluster volume replace-brick mirror `hostname`:/exports/exp5 `hostname`:/exports/exp1 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick mirror `hostname`:/exports/exp5 `hostname`:/exports/exp1 status +gluster volume replace-brick mirror `hostname`:/exports/exp5 `hostname`:/exports/exp1 abort + +gluster volume info +sleep 1 +df -h +sleep 1 + +echo "removeing-brick....." +gluster --mode=script volume remove-brick mirror `hostname`:/exports/exp3 `hostname`:/exports/exp4 +gluster volume info +sleep 1 +df -h +sleep 1 + +echo "stopping replicate volume....." +gluster --mode=script volume stop mirror +gluster volume info +sleep 1 +umount /mnt/client +df -h + +echo "deleting replicate volume....." +gluster --mode=script volume delete mirror +gluster volume info +sleep 1 + +# create stripe volume and try start, mount, add-brick, replace-brick, remove-brick, stop, unmount, delete + +echo "creating stripe volume....." +gluster volume create str stripe 2 `hostname`:/exports/exp1 `hostname`:/exports/exp2 +gluster volume info +sleep 1 + +echo "starting stripe volume....." +gluster volume start str +gluster volume info +sleep 1 +mount -t glusterfs `hostname`:str /mnt/client +sleep 1 +df -h +sleep 1 + +echo "adding brick...." +gluster volume add-brick str `hostname`:/exports/exp3 `hostname`:/exports/exp4 +gluster volume info +sleep 1 +df -h +sleep 1 + +echo "replacing brick....." +gluster volume replace-brick str `hostname`:/exports/exp1 `hostname`:/exports/exp5 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick str `hostname`:/exports/exp1 `hostname`:/exports/exp5 status +gluster volume replace-brick str `hostname`:/exports/exp1 `hostname`:/exports/exp5 pause +gluster volume replace-brick str `hostname`:/exports/exp1 `hostname`:/exports/exp5 status +gluster volume replace-brick str `hostname`:/exports/exp1 `hostname`:/exports/exp5 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick str `hostname`:/exports/exp1 `hostname`:/exports/exp5 status +gluster volume replace-brick str `hostname`:/exports/exp1 `hostname`:/exports/exp5 commit + +gluster volume info +sleep 1 +df -h +sleep 1 + +echo "replacing brick for abort operation" +gluster volume replace-brick str `hostname`:/exports/exp5 `hostname`:/exports/exp1 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick str `hostname`:/exports/exp5 `hostname`:/exports/exp1 status +gluster volume replace-brick str `hostname`:/exports/exp5 `hostname`:/exports/exp1 pause +gluster volume replace-brick str `hostname`:/exports/exp5 `hostname`:/exports/exp1 status +gluster volume replace-brick str `hostname`:/exports/exp5 `hostname`:/exports/exp1 start +#sleep for 5 seconds +sleep 5 +gluster volume replace-brick str `hostname`:/exports/exp5 `hostname`:/exports/exp1 status +gluster volume replace-brick str `hostname`:/exports/exp5 `hostname`:/exports/exp1 abort + +gluster volume info +sleep 1 +df -h +sleep 1 + +echo "removing-brick....." +gluster --mode=script volume remove-brick str `hostname`:/exports/exp3 `hostname`:/exports/exp4 +gluster volume info +sleep 1 +df -h +sleep 1 + +echo "stopping stripe volume....." +gluster --mode=script volume stop str +gluster volume info +sleep 1 +umount /mnt/client +df -h + +echo "deleting stripe volume....." +gluster --mode=script volume delete str +gluster volume info + diff --git a/extras/test/ld-preload-test/Makefile b/extras/test/ld-preload-test/Makefile new file mode 100644 index 000000000..97aa01651 --- /dev/null +++ b/extras/test/ld-preload-test/Makefile @@ -0,0 +1,28 @@ +TOOL_CFLAGS32=-g -O0 -Wall +TOOL_CFLAGS64=-g -O0 -D_GNU_SOURCE -Wall -D_FILE_OFFSET_BITS=64 +LIB_CFLAGS=-g -O0 -fPIC -Wall -shared -ldl +%LIB_CFLAGS=-g -O0 -D_GNU_SOURCE -fPIC -Wall -shared -D_FILE_OFFSET_BITS=64 -ldl + +all: ld-preload-test32 ld-preload-test64 ld-preload-lib + + +copy32: + cp ld-preload-test.c test32.c + +ld-preload-test32: copy32 + gcc ${TOOL_CFLAGS32} test32.c -o ld-preload-test32 + rm -f test32.c + +copy64: + cp ld-preload-test.c test64.c + +ld-preload-test64: copy64 + gcc ${TOOL_CFLAGS64} test64.c -o ld-preload-test64 + rm -f test64.c + +ld-preload-lib: ld-preload-lib.c + gcc ${LIB_CFLAGS} ld-preload-lib.c -o ld-preload-lib.so + +clean: + rm -f ld-preload-test32 ld-preload-test64 ld-preload-lib.so + diff --git a/extras/test/ld-preload-test/README b/extras/test/ld-preload-test/README new file mode 100644 index 000000000..725b94023 --- /dev/null +++ b/extras/test/ld-preload-test/README @@ -0,0 +1,40 @@ +ld-preload-test +=============== +1. The idea of this test tool is to check the sanity of the LD_PRELOAD +mechanism in libc for the system on which booster needs to run. + +2. Basically, this test tool calls various system calls for which there is +support in the booster library. Combined with the logging output from this +tool and the preload-test-lib, one can determine whether the +pre-loading mechanism is working working in order for booster to initialize. + +3. This tool does not test GlusterFS functionality running under booster +although the path specified to the tool can be a GlusterFS mountpoint but +that is not very useful. + +4. This tool is incomplete without the preload-test-lib and the +accompanyung shell script that needs to be run for running the test. + +ld-preload-lib.so +================= +A very simple library that intercepts booster supported system calls +and prints a log message to stdout. +Combined with the ld-preload-test, we cam determine whether all system calls +are getting redirected into this library when LD_PRELOAD'ed. This helps us +conduct a basic test to ensure that the required syscalls actually will +be intercepted by the booster library. + + +Instructions +============ + +1. Build the binaries. + $ make + +(We've tested the tool on Debian and CentOS. If there are build errors or +warnings, please do report them to glusterfs-devel@nongnu.org.) + +2. Run the test. + $ ./test-preload.sh > preload.log + +3. Mail the log to glusterfs-devel@nongnu.org. diff --git a/extras/test/ld-preload-test/ld-preload-lib.c b/extras/test/ld-preload-test/ld-preload-lib.c new file mode 100644 index 000000000..88afd14c3 --- /dev/null +++ b/extras/test/ld-preload-test/ld-preload-lib.c @@ -0,0 +1,627 @@ +/* + Copyright (c) 2007-2012 Red Hat, Inc. <http://www.redhat.com> + This file is part of GlusterFS. + + This file is licensed to you under your choice of the GNU Lesser + General Public License, version 3 or any later version (LGPLv3 or + later), or the GNU General Public License, version 2 (GPLv2), in all + cases as published by the Free Software Foundation. +*/ + +/* LD PRELOAD'able library + * A very simple library that intercepts booster supported system calls + * and prints a log message to stdout. + * + * Combined with the ld-preload-test, we cam determine whether all system calls + * are getting redirected into this library when LD_PRELOAD'ed. This helps us + * conduct a basic test to ensure that the required syscalls actually will + * be intercepted by the booster library. + */ + +#include <dlfcn.h> +#include <sys/types.h> +#include <sys/uio.h> +#include <stdio.h> +#include <stdarg.h> +#include <stdlib.h> +#include <inttypes.h> +#include <string.h> +#include <assert.h> +#include <errno.h> +#include <utime.h> +#include <sys/statfs.h> +#include <sys/statvfs.h> +#include <fcntl.h> +#include <sys/stat.h> +#include <dirent.h> +#include <attr/xattr.h> +#include <sys/sendfile.h> + +/* Err number that is assigned to errno so that test application can + * verify that the function was intercepted correctly. + */ +#define PRELOAD_ERRNO_VERF 6449 +#define set_errno() (errno = PRELOAD_ERRNO_VERF) + +inline void +intercept (char *call, int tabs) +{ + while (tabs > 0) { + fprintf (stdout, "\t"); + --tabs; + } + + fprintf (stdout, "Intercepted by %s", call); +} + +int +creat64 (const char *pathname, mode_t mode) +{ + intercept ("creat64", 2); + set_errno (); + return -1; +} + +int +creat (const char *pathname, mode_t mode) +{ + intercept ("creat", 2); + set_errno (); + return -1; +} + + +int +close (int fd) +{ + intercept ("close", 2); + set_errno (); + return -1; +} + +int +open64 (const char *pathname, int flags, ...) +{ + intercept ("open64", 2); + set_errno (); + return -1; +} + + +int +open (const char *pathname, int flags, ...) +{ + intercept ("open", 2); + set_errno (); + return -1; +} + +ssize_t +read (int fd, void *buf, size_t count) +{ + intercept ("read", 2); + set_errno (); + return -1; +} + +ssize_t +readv (int fd, const struct iovec *vector, int count) +{ + intercept ("readv", 2); + set_errno (); + return -1; +} + +ssize_t +pread (int fd, void *buf, size_t count, unsigned long offset) +{ + intercept ("pread", 2); + set_errno (); + return -1; +} + + +ssize_t +pread64 (int fd, void *buf, size_t count, uint64_t offset) +{ + intercept ("pread64", 2); + set_errno (); + return -1; +} + +ssize_t +write (int fd, const void *buf, size_t count) +{ + intercept ("write", 2); + set_errno (); + return -1; +} + +ssize_t +writev (int fd, const struct iovec *vector, int count) +{ + intercept ("writev", 2); + set_errno (); + return -1; +} + +ssize_t +pwrite (int fd, const void *buf, size_t count, unsigned long offset) +{ + intercept ("pwrite", 2); + set_errno (); + return -1; +} + +ssize_t +pwrite64 (int fd, const void *buf, size_t count, uint64_t offset) +{ + intercept ("pwrite64", 2); + set_errno (); + return -1; +} + + +off_t +lseek (int fildes, unsigned long offset, int whence) +{ + intercept ("lseek", 2); + set_errno (); + return -1; +} + +off_t +lseek64 (int fildes, uint64_t offset, int whence) +{ + intercept ("lseek64", 2); + set_errno (); + return -1; +} + + +int +dup (int fd) +{ + intercept ("dup", 2); + set_errno (); + return -1; +} + +int +dup2 (int oldfd, int newfd) +{ + intercept ("dup2", 2); + set_errno (); + return -1; +} + +int +mkdir (const char *pathname, mode_t mode) +{ + intercept ("mkdir", 2); + set_errno (); + return -1; +} + +int +rmdir (const char *pathname) +{ + intercept ("rmdir", 2); + set_errno (); + return -1; +} + +int +chmod (const char *pathname, mode_t mode) +{ + intercept ("chmod", 2); + set_errno (); + return -1; +} + +int +chown (const char *pathname, uid_t owner, gid_t group) +{ + intercept ("chown", 2); + set_errno (); + return -1; +} + +int +fchmod (int fd, mode_t mode) +{ + intercept ("fchmod", 2); + set_errno (); + return -1; +} + +int +fchown (int fd, uid_t uid, gid_t gid) +{ + intercept ("fchown", 2); + set_errno (); + return -1; +} + +int fsync (int fd) +{ + intercept ("fsync", 2); + set_errno (); + return -1; +} + + +int +ftruncate (int fd, off_t length) +{ + intercept ("ftruncate", 1); + set_errno (); + return -1; +} + + +int +ftruncate64 (int fd, off_t length) +{ + intercept ("ftruncate64", 1); + set_errno (); + return -1; +} + +int +link (const char *oldpath, const char *newname) +{ + intercept ("link", 2); + set_errno (); + return -1; +} + +int +rename (const char *oldpath, const char *newpath) +{ + intercept ("rename", 2); + set_errno (); + return -1; +} + +int +utimes (const char *path, const struct timeval times[2]) +{ + intercept ("utimes", 2); + set_errno (); + return -1; +} + +int +utime (const char *path, const struct utimbuf *buf) +{ + intercept ("utime", 2); + set_errno (); + return -1; +} + + +int +mknod (const char *path, mode_t mode, dev_t dev) +{ + intercept ("mknod", 2); + set_errno (); + return -1; +} + +int +__xmknod (int ver, const char *path, mode_t mode, dev_t *dev) +{ + intercept ("__xmknod", 2); + set_errno (); + return -1; +} + +int +mkfifo (const char *path, mode_t mode) +{ + intercept ("mkfifo", 2); + set_errno (); + return -1; +} + +int +unlink (const char *path) +{ + intercept ("unlink", 2); + set_errno (); + return -1; +} + + +int +symlink (const char *oldpath, const char *newpath) +{ + intercept ("symlink", 2); + set_errno (); + return -1; +} + +int +readlink (const char *path, char *buf, size_t bufsize) +{ + intercept ("readlink", 1); + set_errno (); + return -1; +} + + +char * +realpath (const char *path, char *resolved) +{ + intercept ("realpath", 1); + set_errno (); + return NULL; +} + + +DIR * +opendir (const char *path) +{ + intercept ("opendir", 2); + set_errno (); + return NULL; +} + + +struct dirent * +readdir (DIR *dir) +{ + intercept ("readdir\t", 2); + set_errno (); + return NULL; +} + +struct dirent * +readdir64 (DIR *dir) +{ + intercept ("readdir64", 2); + set_errno (); + return NULL; +} + + +int +readdir_r (DIR *dir, struct dirent *entry, struct dirent **result) +{ + intercept ("readdir_r", 1); + set_errno (); + return -1; +} + +int +readdir64_r (DIR *dir, struct dirent *entry, struct dirent **result) +{ + intercept ("readdir64_r", 1); + set_errno (); + return -1; +} + + +int +closedir (DIR *dh) +{ + intercept ("closedir", 1); + set_errno (); + return -1; +} + +int +__xstat (int ver, const char *path, struct stat *buf) +{ + intercept ("__xstat\t", 2); + set_errno (); + return -1; +} + + +int +__xstat64 (int ver, const char *path, struct stat *buf) +{ + intercept ("__xstat64", 2); + set_errno (); + return -1; +} + +int +stat (const char *path, struct stat *buf) +{ + intercept ("stat", 2); + set_errno (); + return -1; +} + +int +stat64 (const char *path, struct stat *buf) +{ + intercept ("stat64", 2); + set_errno (); + return -1; +} + +int +__fxstat (int ver, int fd, struct stat *buf) +{ + intercept ("__fxstat\t", 2); + set_errno (); + return -1; +} + + +int +__fxstat64 (int ver, int fd, struct stat *buf) +{ + intercept ("__fxstat64", 2); + set_errno (); + return -1; +} + +int +fstat (int fd, struct stat *buf) +{ + intercept ("fstat", 2); + set_errno (); + return -1; +} + +int +fstat64 (int fd , struct stat *buf) +{ + intercept ("fstat64", 2); + set_errno (); + return -1; +} + +int +__lxstat (int ver, const char *path, struct stat *buf) +{ + intercept ("__lxstat\t", 2); + set_errno (); + return -1; +} + +int +__lxstat64 (int ver, const char *path, struct stat *buf) +{ + intercept ("__lxstat64", 2); + set_errno (); + return -1; +} + +int +lstat (const char *path, struct stat *buf) +{ + intercept ("lstat", 2); + set_errno (); + return -1; +} + +int +lstat64 (const char *path, struct stat *buf) +{ + intercept ("lstat64", 2); + set_errno (); + return -1; +} + +int +statfs (const char *path, struct statfs *buf) +{ + intercept ("statfs", 2); + set_errno (); + return -1; +} + + +int +statfs64 (const char *path, struct statfs *buf) +{ + intercept ("statfs64", 2); + set_errno (); + return -1; +} + +int +statvfs (const char *path, struct statvfs *buf) +{ + intercept ("statvfs\t", 2); + set_errno (); + return -1; +} + + +int +statvfs64 (const char *path, struct statvfs *buf) +{ + intercept ("statvfs64", 2); + set_errno (); + return -1; +} + +ssize_t +getxattr (const char *path, const char *name, void *value, size_t size) +{ + intercept ("getxattr", 1); + set_errno (); + return -1; +} + +ssize_t +lgetxattr (const char *path, const char *name, void *value, size_t size) +{ + intercept ("lgetxattr", 1); + set_errno (); + return -1; +} + + +int +remove (const char* path) +{ + intercept ("remove", 2); + set_errno (); + return -1; +} + +int +lchown (const char *path, uid_t owner, gid_t group) +{ + intercept ("lchown", 2); + set_errno (); + return -1; +} + +void +rewinddir (DIR *dirp) +{ + intercept ("rewinddir", 1); + set_errno (); + return; +} + +void +seekdir (DIR *dirp, off_t offset) +{ + intercept ("seekdir", 2); + set_errno (); + return; +} + +off_t +telldir (DIR *dirp) +{ + intercept ("telldir", 2); + set_errno (); + return -1; +} + +ssize_t +sendfile (int out_fd, int in_fd, off_t *offset, size_t count) +{ + intercept ("sendfile\t", 1); + set_errno (); + return -1; +} + +ssize_t +sendfile64 (int out_fd, int in_fd, off_t *offset, size_t count) +{ + intercept ("sendfile64", 1); + set_errno (); + return -1; +} + + +int +fcntl (int fd, int cmd, ...) +{ + intercept ("fcntl", 2); + set_errno (); + return -1; +} + diff --git a/extras/test/ld-preload-test/ld-preload-test.c b/extras/test/ld-preload-test/ld-preload-test.c new file mode 100644 index 000000000..78772f598 --- /dev/null +++ b/extras/test/ld-preload-test/ld-preload-test.c @@ -0,0 +1,367 @@ +/* + Copyright (c) 2007-2012 Red Hat, Inc. <http://www.redhat.com> + This file is part of GlusterFS. + + This file is licensed to you under your choice of the GNU Lesser + General Public License, version 3 or any later version (LGPLv3 or + later), or the GNU General Public License, version 2 (GPLv2), in all + cases as published by the Free Software Foundation. +*/ + +/* + * LD PRELOAD Test Tool + * + * 1. The idea of this test tool is to check the sanity of the LD_PRELOAD + * mechanism in libc for the system on which booster needs to run. + * + * 2. Basically, this test tool calls various system calls for which there is + * support in the booster library. Combined with the logging output from this + * tool and the preload-test-lib, one can determine whether the + * pre-loading mechanism is working working in order for booster to initialize. + * + * 3. This tool does not test GlusterFS functionality running under booster + * although the path specified to the tool can be a GlusterFS mountpoint but + * that is not very useful. + * + * 4. This tool is incomplete without the preload-test-lib and the + * accompanyung shell script that needs to be run for running the test. + */ +#define _XOPEN_SOURCE 500 + +#include <dlfcn.h> +#include <sys/types.h> +#include <stdio.h> +#include <stdarg.h> +#include <stdlib.h> +#include <inttypes.h> +#include <string.h> +#include <assert.h> +#include <errno.h> +#include <dirent.h> +#include <sys/statfs.h> +#include <sys/statvfs.h> +#include <sys/stat.h> +#include <fcntl.h> +#include <unistd.h> +#include <sys/uio.h> +#include <utime.h> +#include <sys/time.h> +#include <attr/xattr.h> +#include <sys/sendfile.h> + + +#define PRELOAD_ERRNO_VERF 6449 +inline void +check_err(int ret, char *call, int tabs) +{ + while (tabs > 0) { + fprintf (stdout, "\t"); + --tabs; + } + if (ret != -1) { + fprintf (stdout, "Not intercepted: %s\n", call); + return; + } + + if (errno != PRELOAD_ERRNO_VERF) { + fprintf (stdout, "Not intercepted: %s: err: %s\n", call, + strerror (errno)); + return; + } + + fprintf (stdout, "Intercept verified: %s\n", call); + return; +} + +void +usage (FILE *fp) +{ + fprintf (fp, "Usage: ld-preload-test <Options>\n"); + fprintf (fp, "Options\n"); + fprintf (fp, "\t--path\t\tPathname is used as the file/directory" + " created for the test.\n"); + +} + + +int +run_file_tests (char *testfile) +{ + int ret = -1; + struct stat buf; + + assert (testfile); + fprintf (stdout, "Testing creat"); + ret = creat (testfile, S_IRWXU); + check_err (ret, "creat", 2); + + fprintf (stdout, "Testing close"); + ret = close (ret); + check_err (ret, "close", 2); + + fprintf (stdout, "Testing open"); + ret = open (testfile, O_RDONLY); + check_err (ret, "open", 2); + + fprintf (stdout, "Testing read"); + ret = read (0, NULL, 0); + check_err (ret, "read", 2); + + fprintf (stdout, "Testing readv"); + ret = readv (0, NULL, 0); + check_err (ret, "readv", 2); + + fprintf (stdout, "Testing pread"); + ret = pread (0, NULL, 0, 0); + check_err (ret, "pread", 2); + + fprintf (stdout, "Testing write"); + ret = write (0, NULL, 0); + check_err (ret, "write", 2); + + fprintf (stdout, "Testing writev"); + ret = writev (0, NULL, 0); + check_err (ret, "writev", 2); + + fprintf (stdout, "Testing pwrite"); + ret = pwrite (0, NULL, 0, 0); + check_err (ret, "pwrite", 2); + + fprintf (stdout, "Testing lseek"); + ret = lseek (0, 0, 0); + check_err (ret, "lseek", 2); + + fprintf (stdout, "Testing dup"); + ret = dup (0); + check_err (ret, "dup", 2); + + fprintf (stdout, "Testing dup2"); + ret = dup2 (0, 0); + check_err (ret, "dup2", 2); + + fprintf (stdout, "Testing fchmod"); + ret = fchmod (0, 0); + check_err (ret, "fchmod", 2); + + fprintf (stdout, "Testing fchown"); + ret = fchown (0, 0, 0); + check_err (ret, "fchown", 2); + + fprintf (stdout, "Testing fsync"); + ret = fsync (0); + check_err (ret, "fsync", 2); + + fprintf (stdout, "Testing ftruncate"); + ret = ftruncate (0, 0); + check_err (ret, "ftruncate", 1); + + fprintf (stdout, "Testing fstat"); + ret = fstat (0, &buf); + check_err (ret, "fstat", 1); + + fprintf (stdout, "Testing sendfile"); + ret = sendfile (0, 0, NULL, 0); + check_err (ret, "sendfile", 1); + + fprintf (stdout, "Testing fcntl"); + ret = fcntl (0, 0, NULL); + check_err (ret, "fcntl", 2); + + fprintf (stdout, "Testing close"); + ret = close (ret); + check_err (ret, "close", 2); + + fprintf (stdout, "Testing remove"); + ret = remove (testfile); + check_err (ret, "remove", 2); + + return ret; +} + + +int +run_attr_tests (char *testfile) +{ + int ret = -1; + char *res = NULL; + struct stat buf; + struct statfs sbuf; + struct statvfs svbuf; + + assert (testfile); + + fprintf (stdout, "Testing chmod"); + ret = chmod (testfile, 0); + check_err (ret, "chmod", 2); + + fprintf (stdout, "Testing chown"); + ret = chown (testfile, 0, 0); + check_err (ret, "chown", 2); + + fprintf (stdout, "Testing link"); + ret = link (testfile, testfile); + check_err (ret, "link", 2); + + fprintf (stdout, "Testing rename"); + ret = rename (testfile, testfile); + check_err (ret, "rename", 2); + + fprintf (stdout, "Testing utimes"); + ret = utimes (testfile, NULL); + check_err (ret, "utimes", 2); + + fprintf (stdout, "Testing utime"); + ret = utime (testfile, NULL); + check_err (ret, "utime", 2); + + fprintf (stdout, "Testing unlink"); + ret = unlink (testfile); + check_err (ret, "unlink", 2); + + fprintf (stdout, "Testing symlink"); + ret = symlink (testfile, testfile); + check_err (ret, "symlink", 2); + + fprintf (stdout, "Testing readlink"); + ret = readlink (testfile, testfile, 0); + check_err (ret, "readlink", 2); + + fprintf (stdout, "Testing realpath"); + ret = 0; + res = realpath ((const char *)testfile, testfile); + if (!res) + ret = -1; + check_err (ret, "realpath", 2); + + fprintf (stdout, "Testing stat"); + ret = stat (testfile, &buf); + check_err (ret, "stat", 1); + + fprintf (stdout, "Testing lstat"); + ret = lstat (testfile, &buf); + check_err (ret, "lstat", 1); + + fprintf (stdout, "Testing statfs"); + ret = statfs (testfile, &sbuf); + check_err (ret, "statfs", 2); + + fprintf (stdout, "Testing statvfs"); + ret = statvfs (testfile, &svbuf); + check_err (ret, "statvfs", 1); + + fprintf (stdout, "Testing getxattr"); + ret = getxattr (testfile, NULL, NULL, 0); + check_err (ret, "getxattr", 2); + + fprintf (stdout, "Testing lgetxattr"); + ret = lgetxattr (testfile, NULL, NULL, 0); + check_err (ret, "lgetxattr", 1); + + fprintf (stdout, "Testing lchown"); + ret = lchown (testfile, 0, 0); + check_err (ret, "lchown", 2); + return 0; +} + + +int +run_dev_tests (char *testfile) +{ + int ret = -1; + + assert (testfile); + + fprintf (stdout, "Testing mknod"); + ret = mknod (testfile, 0, 0); + check_err (ret, "mknod", 2); + + fprintf (stdout, "Testing mkfifo"); + ret = mkfifo (testfile, 0); + check_err (ret, "mkfifo", 2); + return 0; +} + +int +run_dir_tests (char *testpath) +{ + int ret = -1; + DIR *dh = NULL; + struct dirent *dire = NULL; + + assert (testpath); + + fprintf (stdout, "Testing mkdir"); + ret = mkdir (testpath, 0); + check_err (ret, "mkdir", 2); + + fprintf (stdout, "Testing rmdir"); + ret = rmdir (testpath); + check_err (ret, "rmdir", 2); + + fprintf (stdout, "Testing opendir"); + ret = 0; + dh = opendir (testpath); + if (!dh) + ret = -1; + check_err (ret, "opendir", 2); + + fprintf (stdout, "Testing readdir"); + ret = 0; + dire = readdir (dh); + if (!dire) + ret = -1; + check_err (ret, "readdir", 1); + + fprintf (stdout, "Testing readdir_r"); + ret = readdir_r (dh, dire, &dire); + check_err (ret, "readdir_r", 1); + + fprintf (stdout, "Testing rewinddir"); + rewinddir (dh); + check_err (-1, "rewinddir", 1); + + fprintf (stdout, "Testing seekdir"); + seekdir (dh, 0); + check_err (-1, "seekdir", 2); + + fprintf (stdout, "Testing telldir"); + ret = telldir (dh); + check_err (ret, "telldir", 2); + + fprintf (stdout, "Testing closedir"); + ret = closedir (dh); + check_err (ret, "closedir", 2); + return 0; +} + + + +int +main (int argc, char *argv[]) +{ + char *testpath = NULL; + int x = 0; + + for (;x < argc; ++x) { + if (strcmp (argv[x], "--path") == 0) { + testpath = argv[x+1]; + continue; + } + + } + + if (!testpath) { + fprintf (stderr, "--path not specified\n"); + usage (stderr); + return -1; + } + + run_file_tests (testpath); + run_dir_tests (testpath); + run_attr_tests (testpath); + run_dev_tests (testpath); + + return 0; +} + + diff --git a/extras/test/ld-preload-test/test-preload.sh b/extras/test/ld-preload-test/test-preload.sh new file mode 100755 index 000000000..faed5cee3 --- /dev/null +++ b/extras/test/ld-preload-test/test-preload.sh @@ -0,0 +1,7 @@ +#!/bin/sh + +echo "Testing 32-bit offsets build" +LD_PRELOAD="$(pwd)/ld-preload-lib.so" $(pwd)/ld-preload-test32 --path /tmp/testfile +echo +echo "Testing 64-bit offsets build" +LD_PRELOAD="$(pwd)/ld-preload-lib.so" $(pwd)/ld-preload-test64 --path /tmp/testfile diff --git a/extras/test/open-fd-tests.c b/extras/test/open-fd-tests.c new file mode 100644 index 000000000..4184079d0 --- /dev/null +++ b/extras/test/open-fd-tests.c @@ -0,0 +1,64 @@ + +#include <stdio.h> +#include <fcntl.h> +#include <unistd.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <attr/xattr.h> +#include <errno.h> +#include <string.h> + +int +main (int argc, char *argv[]) +{ + int ret = -1; + int fd = 0; + char *filename = NULL; + int loop = 0; + struct stat stbuf = {0,}; + char string[1024] = {0,}; + + if (argc > 1) + filename = argv[1]; + + if (!filename) + filename = "temp-fd-test-file"; + + fd = open (filename, O_RDWR|O_CREAT|O_TRUNC); + if (fd < 0) { + fd = 0; + fprintf (stderr, "open failed : %s\n", strerror (errno)); + goto out; + } + + while (loop < 1000) { + /* Use it as a mechanism to test time delays */ + memset (string, 0, 1024); + scanf ("%s", string); + + ret = write (fd, string, strlen (string)); + if (ret != strlen (string)) { + fprintf (stderr, "write failed : %s (%s %d)\n", + strerror (errno), string, loop); + goto out; + } + + ret = write (fd, "\n", 1); + if (ret != 1) { + fprintf (stderr, "write failed : %s (%d)\n", + strerror (errno), loop); + goto out; + } + + loop++; + } + + fprintf (stdout, "finishing the test after %d loops\n", loop); + + ret = 0; +out: + if (fd) + close (fd); + + return ret; +} diff --git a/extras/test/run.sh b/extras/test/run.sh new file mode 100755 index 000000000..4b3839cf6 --- /dev/null +++ b/extras/test/run.sh @@ -0,0 +1,22 @@ +#!/bin/sh + +# Copyright (c) 2006-2012 Red Hat, Inc. <http://www.redhat.com> +# This file is part of GlusterFS. +# +# This file is licensed to you under your choice of the GNU Lesser +# General Public License, version 3 or any later version (LGPLv3 or +# later), or the GNU General Public License, version 2 (GPLv2), in all +# cases as published by the Free Software Foundation. + +# Running gluster sanity test which starts glusterd and runs gluster commands, and exit at the first failure. +$PWD/gluster_commands.sh + +if [ $? -ne 0 ]; then + echo "sanity failed" +else + echo "sanity passed" +fi + +# Stopping glusterd +$PWD/stop_glusterd.sh + diff --git a/extras/test/stop_glusterd.sh b/extras/test/stop_glusterd.sh new file mode 100755 index 000000000..a2db13f42 --- /dev/null +++ b/extras/test/stop_glusterd.sh @@ -0,0 +1,19 @@ +#!/bin/bash + +# Copyright (c) 2006-2012 Red Hat, Inc. <http://www.redhat.com> +# This file is part of GlusterFS. +# +# This file is licensed to you under your choice of the GNU Lesser +# General Public License, version 3 or any later version (LGPLv3 or +# later), or the GNU General Public License, version 2 (GPLv2), in all +# cases as published by the Free Software Foundation. + +#This script stops the glusterd running on the machine. Helpful for gluster sanity script + +killall -9 glusterd + +if [ $? -ne 0 ]; then + echo "Error: Could not kill glusterd. Either glusterd is not running or kill it manually" +else + echo "Killed glusterd" +fi diff --git a/extras/test/test-ffop.c b/extras/test/test-ffop.c new file mode 100644 index 000000000..2d174d452 --- /dev/null +++ b/extras/test/test-ffop.c @@ -0,0 +1,870 @@ +#include <stdio.h> +#include <fcntl.h> +#include <unistd.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <attr/xattr.h> +#include <errno.h> +#include <string.h> +#include <dirent.h> + +int fd_based_fops_1 (char *filename); //for fd based fops after unlink +int fd_based_fops_2 (char *filename); //for fd based fops before unlink +int dup_fd_based_fops (char *filename); // fops based on fd after dup +int path_based_fops (char *filename); //for fops based on path +int dir_based_fops (char *filename); // for fops which operate on directory +int link_based_fops (char *filename); //for fops which operate in link files (symlinks) +int test_open_modes (char *filename); // to test open syscall with open modes available. +int generic_open_read_write (char *filename, int flag); // generic function which does open write and read. + +int +main (int argc, char *argv[]) +{ + int ret = -1; + char filename[255] = {0,}; + + if (argc > 1) + strcpy(filename, argv[1]); + else + strcpy(filename, "temp-xattr-test-file"); + + ret = fd_based_fops_1 (strcat(filename, "_1")); + if (ret < 0) + fprintf (stderr, "fd based file operation 1 failed\n"); + else + fprintf (stdout, "fd based file operation 1 passed\n"); + + ret = fd_based_fops_2 (strcat(filename, "_2")); + if (ret < 0) + fprintf (stderr, "fd based file operation 2 failed\n"); + else + fprintf (stdout, "fd based file operation 2 passed\n"); + + ret = dup_fd_based_fops (strcat (filename, "_3")); + if (ret < 0) + fprintf (stderr, "dup fd based file operation failed\n"); + else + fprintf (stdout, "dup fd based file operation passed\n"); + + ret = path_based_fops (strcat (filename, "_4")); + if (ret < 0) + fprintf (stderr, "path based file operation failed\n"); + else + fprintf (stdout, "path based file operation passed\n"); + + ret = dir_based_fops (strcat (filename, "_5")); + if (ret < 0) + fprintf (stderr, "directory based file operation failed\n"); + else + fprintf (stdout, "directory based file operation passed\n"); + + ret = link_based_fops (strcat (filename, "_5")); + if (ret < 0) + fprintf (stderr, "link based file operation failed\n"); + else + fprintf (stdout, "link based file operation passed\n"); + + ret = test_open_modes (strcat (filename, "_5")); + if (ret < 0) + fprintf (stderr, "testing modes of 'open' call failed\n"); + else + fprintf (stdout, "testing modes of 'open' call passed\n"); + +out: + return ret; +} + +int +fd_based_fops_1 (char *filename) +{ + int fd = 0; + int ret = -1; + struct stat stbuf = {0,}; + char wstr[50] = {0,}; + char rstr[50] = {0,}; + + fd = open (filename, O_RDWR|O_CREAT); + if (fd < 0) { + fd = 0; + fprintf (stderr, "open failed : %s\n", strerror (errno)); + goto out; + } + + ret = unlink (filename); + if (ret < 0) { + fprintf (stderr, "unlink failed : %s\n", strerror (errno)); + goto out; + } + + strcpy (wstr, "This is my string\n"); + ret = write (fd, wstr, strlen(wstr)); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "write failed: %s\n", strerror (errno)); + goto out; + } + + ret = lseek (fd, 0, SEEK_SET); + if (ret < 0) { + fprintf (stderr, "lseek failed: %s\n", strerror (errno)); + goto out; + } + + ret = read (fd, rstr, strlen(wstr)); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "read failed: %s\n", strerror (errno)); + goto out; + } + + ret = memcmp (rstr, wstr, strlen (wstr)); + if (ret != 0) { + ret = -1; + fprintf (stderr, "read returning junk\n"); + goto out; + } + + ret = ftruncate (fd, 0); + if (ret < 0) { + fprintf (stderr, "ftruncate failed : %s\n", strerror (errno)); + goto out; + } + + ret = fstat (fd, &stbuf); + if (ret < 0) { + fprintf (stderr, "fstat failed : %s\n", strerror (errno)); + goto out; + } + + ret = fchmod (fd, 0640); + if (ret < 0) { + fprintf (stderr, "fchmod failed : %s\n", strerror (errno)); + goto out; + } + + ret = fchown (fd, 10001, 10001); + if (ret < 0) { + fprintf (stderr, "fchown failed : %s\n", strerror (errno)); + goto out; + } + + ret = fsync (fd); + if (ret < 0) { + fprintf (stderr, "fsync failed : %s\n", strerror (errno)); + goto out; + } + + ret = fsetxattr (fd, "trusted.xattr-test", "working", 8, 0); + if (ret < 0) { + fprintf (stderr, "fsetxattr failed : %s\n", strerror (errno)); + goto out; + } + + ret = fdatasync (fd); + if (ret < 0) { + fprintf (stderr, "fdatasync failed : %s\n", strerror (errno)); + goto out; + } + + ret = flistxattr (fd, NULL, 0); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "flistxattr failed : %s\n", strerror (errno)); + goto out; + } + + ret = fgetxattr (fd, "trusted.xattr-test", NULL, 0); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "fgetxattr failed : %s\n", strerror (errno)); + goto out; + } + + ret = fremovexattr (fd, "trusted.xattr-test"); + if (ret < 0) { + fprintf (stderr, "fremovexattr failed : %s\n", strerror (errno)); + goto out; + } + + ret = 0; +out: + if (fd) + close (fd); + + return ret; +} + + +int +fd_based_fops_2 (char *filename) +{ + int fd = 0; + int ret = -1; + struct stat stbuf = {0,}; + char wstr[50] = {0,}; + char rstr[50] = {0,}; + + fd = open (filename, O_RDWR|O_CREAT); + if (fd < 0) { + fd = 0; + fprintf (stderr, "open failed : %s\n", strerror (errno)); + goto out; + } + + ret = ftruncate (fd, 0); + + if (ret < 0) { + fprintf (stderr, "ftruncate failed : %s\n", strerror (errno)); + goto out; + } + + strcpy (wstr, "This is my second string\n"); + ret = write (fd, wstr, strlen (wstr)); + if (ret < 0) { + ret = -1; + fprintf (stderr, "write failed: %s\n", strerror (errno)); + goto out; + } + + lseek (fd, 0, SEEK_SET); + if (ret < 0) { + fprintf (stderr, "lseek failed: %s\n", strerror (errno)); + goto out; + } + + ret = read (fd, rstr, strlen (wstr)); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "read failed: %s\n", strerror (errno)); + goto out; + } + + ret = memcmp (rstr, wstr, strlen (wstr)); + if (ret != 0) { + ret = -1; + fprintf (stderr, "read returning junk\n"); + goto out; + } + + ret = fstat (fd, &stbuf); + if (ret < 0) { + fprintf (stderr, "fstat failed : %s\n", strerror (errno)); + goto out; + } + + ret = fchmod (fd, 0640); + if (ret < 0) { + fprintf (stderr, "fchmod failed : %s\n", strerror (errno)); + goto out; + } + + ret = fchown (fd, 10001, 10001); + if (ret < 0) { + fprintf (stderr, "fchown failed : %s\n", strerror (errno)); + goto out; + } + + ret = fsync (fd); + if (ret < 0) { + fprintf (stderr, "fsync failed : %s\n", strerror (errno)); + goto out; + } + + ret = fsetxattr (fd, "trusted.xattr-test", "working", 8, 0); + if (ret < 0) { + fprintf (stderr, "fsetxattr failed : %s\n", strerror (errno)); + goto out; + } + + ret = fdatasync (fd); + if (ret < 0) { + fprintf (stderr, "fdatasync failed : %s\n", strerror (errno)); + goto out; + } + + ret = flistxattr (fd, NULL, 0); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "flistxattr failed : %s\n", strerror (errno)); + goto out; + } + + ret = fgetxattr (fd, "trusted.xattr-test", NULL, 0); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "fgetxattr failed : %s\n", strerror (errno)); + goto out; + } + + ret = fremovexattr (fd, "trusted.xattr-test"); + if (ret < 0) { + fprintf (stderr, "fremovexattr failed : %s\n", strerror (errno)); + goto out; + } + +out: + if (fd) + close (fd); + unlink (filename); + + return ret; +} + +int +path_based_fops (char *filename) +{ + int ret = -1; + int fd = 0; + struct stat stbuf = {0,}; + char newfilename[255] = {0,}; + + fd = creat (filename, 0644); + if (fd < 0) { + fprintf (stderr, "creat failed: %s\n", strerror (errno)); + goto out; + } + + ret = truncate (filename, 0); + if (ret < 0) { + fprintf (stderr, "truncate failed: %s\n", strerror (errno)); + goto out; + } + + ret = stat (filename, &stbuf); + if (ret < 0) { + fprintf (stderr, "stat failed: %s\n", strerror (errno)); + goto out; + } + + ret = chmod (filename, 0640); + if (ret < 0) { + fprintf (stderr, "chmod failed: %s\n", strerror (errno)); + goto out; + } + + ret = chown (filename, 10001, 10001); + if (ret < 0) { + fprintf (stderr, "chown failed: %s\n", strerror (errno)); + goto out; + } + + ret = setxattr (filename, "trusted.xattr-test", "working", 8, 0); + if (ret < 0) { + fprintf (stderr, "setxattr failed: %s\n", strerror (errno)); + goto out; + } + + ret = listxattr (filename, NULL, 0); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "listxattr failed: %s\n", strerror (errno)); + goto out; + } + + ret = getxattr (filename, "trusted.xattr-test", NULL, 0); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "getxattr failed: %s\n", strerror (errno)); + goto out; + } + + ret = removexattr (filename, "trusted.xattr-test"); + if (ret < 0) { + fprintf (stderr, "removexattr failed: %s\n", strerror (errno)); + goto out; + } + + ret = access (filename, R_OK|W_OK); + if (ret < 0) { + fprintf (stderr, "access failed: %s\n", strerror (errno)); + goto out; + } + + strcpy (newfilename, filename); + strcat(newfilename, "_new"); + ret = rename (filename, newfilename); + if (ret < 0) { + fprintf (stderr, "rename failed: %s\n", strerror (errno)); + goto out; + } + unlink (newfilename); + +out: + if (fd) + close (fd); + + unlink (filename); + return ret; +} + +int +dup_fd_based_fops (char *filename) +{ + int fd = 0; + int newfd = 0; + int ret = -1; + struct stat stbuf = {0,}; + char wstr[50] = {0,}; + char rstr[50] = {0,}; + + fd = open (filename, O_RDWR|O_CREAT); + if (fd < 0) { + fd = 0; + fprintf (stderr, "open failed : %s\n", strerror (errno)); + goto out; + } + + newfd = dup (fd); + if (newfd < 0) { + ret = -1; + fprintf (stderr, "dup failed: %s\n", strerror (errno)); + goto out; + } + + close (fd); + + strcpy (wstr, "This is my string\n"); + ret = write (newfd, wstr, strlen(wstr)); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "write failed: %s\n", strerror (errno)); + goto out; + } + + ret = lseek (newfd, 0, SEEK_SET); + if (ret < 0) { + fprintf (stderr, "lseek failed: %s\n", strerror (errno)); + goto out; + } + + ret = read (newfd, rstr, strlen(wstr)); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "read failed: %s\n", strerror (errno)); + goto out; + } + + ret = memcmp (rstr, wstr, strlen (wstr)); + if (ret != 0) { + ret = -1; + fprintf (stderr, "read returning junk\n"); + goto out; + } + + ret = ftruncate (newfd, 0); + if (ret < 0) { + fprintf (stderr, "ftruncate failed : %s\n", strerror (errno)); + goto out; + } + + ret = fstat (newfd, &stbuf); + if (ret < 0) { + fprintf (stderr, "fstat failed : %s\n", strerror (errno)); + goto out; + } + + ret = fchmod (newfd, 0640); + if (ret < 0) { + fprintf (stderr, "fchmod failed : %s\n", strerror (errno)); + goto out; + } + + ret = fchown (newfd, 10001, 10001); + if (ret < 0) { + fprintf (stderr, "fchown failed : %s\n", strerror (errno)); + goto out; + } + + ret = fsync (newfd); + if (ret < 0) { + fprintf (stderr, "fsync failed : %s\n", strerror (errno)); + goto out; + } + + ret = fsetxattr (newfd, "trusted.xattr-test", "working", 8, 0); + if (ret < 0) { + fprintf (stderr, "fsetxattr failed : %s\n", strerror (errno)); + goto out; + } + + ret = fdatasync (newfd); + if (ret < 0) { + fprintf (stderr, "fdatasync failed : %s\n", strerror (errno)); + goto out; + } + + ret = flistxattr (newfd, NULL, 0); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "flistxattr failed : %s\n", strerror (errno)); + goto out; + } + + ret = fgetxattr (newfd, "trusted.xattr-test", NULL, 0); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "fgetxattr failed : %s\n", strerror (errno)); + goto out; + } + + ret = fremovexattr (newfd, "trusted.xattr-test"); + if (ret < 0) { + fprintf (stderr, "fremovexattr failed : %s\n", strerror (errno)); + goto out; + } + + ret = 0; +out: + if (newfd) + close (newfd); + ret = unlink (filename); + if (ret < 0) { + fprintf (stderr, "unlink failed : %s\n", strerror (errno)); + goto out; + } + + return ret; +} + +int +dir_based_fops (char *dirname) +{ + int ret = -1; + DIR *dp = NULL; + char buff[255] = {0,}; + struct dirent *dbuff = {0,}; + struct stat stbuff = {0,}; + char newdname[255] = {0,}; + char *cwd = NULL; + + ret = mkdir (dirname, 0755); + if (ret < 0) { + fprintf (stderr, "mkdir failed: %s\n", strerror (errno)); + goto out; + } + + dp = opendir (dirname); + if (dp == NULL) { + fprintf (stderr, "opendir failed: %s\n", strerror (errno)); + goto out; + } + + dbuff = readdir (dp); + if (NULL == dbuff) { + fprintf (stderr, "readdir failed: %s\n", strerror (errno)); + goto out; + } + + ret = closedir (dp); + if (ret < 0) { + fprintf (stderr, "closedir failed: %s\n", strerror (errno)); + goto out; + } + + ret = stat (dirname, &stbuff); + if (ret < 0) { + fprintf (stderr, "stat failed: %s\n", strerror (errno)); + goto out; + } + + ret = chmod (dirname, 0744); + if (ret < 0) { + fprintf (stderr, "chmod failed: %s\n", strerror (errno)); + goto out; + } + + ret = chown (dirname, 10001, 10001); + if (ret < 0) { + fprintf (stderr, "chmod failed: %s\n", strerror (errno)); + goto out; + } + + ret = setxattr (dirname, "trusted.xattr-test", "working", 8, 0); + if (ret < 0) { + fprintf (stderr, "setxattr failed: %s\n", strerror (errno)); + goto out; + } + + ret = listxattr (dirname, NULL, 0); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "listxattr failed: %s\n", strerror (errno)); + goto out; + } + + ret = getxattr (dirname, "trusted.xattr-test", NULL, 0); + if (ret <= 0) { + ret = -1; + fprintf (stderr, "getxattr failed: %s\n", strerror (errno)); + goto out; + } + + ret = removexattr (dirname, "trusted.xattr-test"); + if (ret < 0) { + fprintf (stderr, "removexattr failed: %s\n", strerror (errno)); + goto out; + } + + strcpy (newdname, dirname); + strcat (newdname, "/../"); + ret = chdir (newdname); + if (ret < 0) { + fprintf (stderr, "chdir failed: %s\n", strerror (errno)); + goto out; + } + + cwd = getcwd (buff, 255); + if (NULL == cwd) { + fprintf (stderr, "getcwd failed: %s\n", strerror (errno)); + goto out; + } + + strcpy (newdname, dirname); + strcat (newdname, "new"); + ret = rename (dirname, newdname); + if (ret < 0) { + fprintf (stderr, "rename failed: %s\n", strerror (errno)); + goto out; + } + + ret = rmdir (newdname); + if (ret < 0) { + fprintf (stderr, "rmdir failed: %s\n", strerror (errno)); + return ret; + } + +out: + rmdir (dirname); + return ret; +} + +int +link_based_fops (char *filename) +{ + int ret = -1; + int fd = 0; + char newname[255] = {0,}; + char linkname[255] = {0,}; + struct stat lstbuf = {0,}; + + fd = creat (filename, 0644); + if (fd < 0) { + fd = 0; + fprintf (stderr, "creat failed: %s\n", strerror (errno)); + goto out; + } + + strcpy (newname, filename); + strcat (newname, "_hlink"); + ret = link (filename, newname); + if (ret < 0) { + fprintf (stderr, "link failed: %s\n", strerror (errno)); + goto out; + } + + ret = unlink (filename); + if (ret < 0) { + fprintf (stderr, "unlink failed: %s\n", strerror (errno)); + goto out; + } + + strcpy (linkname, filename); + strcat (linkname, "_slink"); + ret = symlink (newname, linkname); + if (ret < 0) { + fprintf (stderr, "symlink failed: %s\n", strerror (errno)); + goto out; + } + + ret = lstat (linkname, &lstbuf); + if (ret < 0) { + fprintf (stderr, "lstbuf failed: %s\n", strerror (errno)); + goto out; + } + + ret = lchown (linkname, 10001, 10001); + if (ret < 0) { + fprintf (stderr, "lchown failed: %s\n", strerror (errno)); + goto out; + } + + ret = lsetxattr (linkname, "trusted.lxattr-test", "working", 8, 0); + if (ret < 0) { + fprintf (stderr, "lsetxattr failed: %s\n", strerror (errno)); + goto out; + } + + ret = llistxattr (linkname, NULL, 0); + if (ret < 0) { + ret = -1; + fprintf (stderr, "llistxattr failed: %s\n", strerror (errno)); + goto out; + } + + ret = lgetxattr (linkname, "trusted.lxattr-test", NULL, 0); + if (ret < 0) { + ret = -1; + fprintf (stderr, "lgetxattr failed: %s\n", strerror (errno)); + goto out; + } + + ret = lremovexattr (linkname, "trusted.lxattr-test"); + if (ret < 0) { + fprintf (stderr, "lremovexattr failed: %s\n", strerror (errno)); + goto out; + } + + +out: + if (fd) + close(fd); + unlink (linkname); + unlink (newname); +} + +int +test_open_modes (char *filename) +{ + int ret = -1; + + ret = generic_open_read_write (filename, O_CREAT|O_WRONLY); + if (3 != ret) { + fprintf (stderr, "flag O_CREAT|O_WRONLY failed: \n"); + goto out; + } + + ret = generic_open_read_write (filename, O_CREAT|O_RDWR); + if (ret != 0) { + fprintf (stderr, "flag O_CREAT|O_RDWR failed\n"); + goto out; + } + + ret = generic_open_read_write (filename, O_CREAT|O_RDONLY); + if (ret != 0) { + fprintf (stderr, "flag O_CREAT|O_RDONLY failed\n"); + goto out; + } + + ret = creat (filename, 0644); + close (ret); + ret = generic_open_read_write (filename, O_WRONLY); + if (3 != ret) { + fprintf (stderr, "flag O_WRONLY failed\n"); + goto out; + } + + ret = creat (filename, 0644); + close (ret); + ret = generic_open_read_write (filename, O_RDWR); + if (0 != ret) { + fprintf (stderr, "flag O_RDWR failed\n"); + goto out; + } + + ret = creat (filename, 0644); + close (ret); + ret = generic_open_read_write (filename, O_RDONLY); + if (0 != ret) { + fprintf (stderr, "flag O_RDONLY failed\n"); + goto out; + } + + ret = creat (filename, 0644); + close (ret); + ret = generic_open_read_write (filename, O_TRUNC|O_WRONLY); + if (3 != ret) { + fprintf (stderr, "flag O_TRUNC|O_WRONLY failed\n"); + goto out; + } + +#if 0 /* undefined behaviour, unable to reliably test */ + ret = creat (filename, 0644); + close (ret); + ret = generic_open_read_write (filename, O_TRUNC|O_RDONLY); + if (0 != ret) { + fprintf (stderr, "flag O_TRUNC|O_RDONLY failed\n"); + goto out; + } +#endif + + ret = generic_open_read_write (filename, O_CREAT|O_RDWR|O_SYNC); + if (0 != ret) { + fprintf (stderr, "flag O_CREAT|O_RDWR|O_SYNC failed\n"); + goto out; + } + + ret = creat (filename, 0644); + close (ret); + ret = generic_open_read_write (filename, O_CREAT|O_EXCL); + if (0 != ret) { + fprintf (stderr, "flag O_CREAT|O_EXCL failed\n"); + goto out; + } + +out: + return ret; +} + +int generic_open_read_write (char *filename, int flag) +{ + int fd = 0; + int ret = -1; + char wstring[50] = {0,}; + char rstring[50] = {0,}; + + fd = open (filename, flag); + if (fd < 0) { + if (flag == O_CREAT|O_EXCL && errno == EEXIST) { + unlink (filename); + return 0; + } + else { + fd = 0; + fprintf (stderr, "open failed: %s\n", strerror (errno)); + return 1; + } + } + + strcpy (wstring, "My string to write\n"); + ret = write (fd, wstring, strlen(wstring)); + if (ret <= 0) { + if (errno != EBADF) { + fprintf (stderr, "write failed: %s\n", strerror (errno)); + close (fd); + unlink(filename); + return 2; + } + } + + ret = lseek (fd, 0, SEEK_SET); + if (ret < 0) { + close (fd); + unlink(filename); + return 4; + } + + ret = read (fd, rstring, strlen(wstring)); + if (ret < 0) { + close (fd); + unlink (filename); + return 3; + } + + /* Compare the rstring with wstring. But we do not want to return + * error when the flag is either O_RDONLY, O_CREAT|O_RDONLY or + * O_TRUNC|O_RDONLY. Because in that case we are not writing + * anything to the file.*/ + + ret = memcmp (wstring, rstring, strlen (wstring)); + if (0 != ret && !(flag == O_CREAT|O_RDONLY || flag == O_RDONLY ||\ + flag == O_TRUNC|O_RDONLY)) { + fprintf (stderr, "read is returning junk\n"); + close (fd); + unlink (filename); + return 4; + } + + close (fd); + unlink (filename); + return 0; +} diff --git a/extras/volfilter.py b/extras/volfilter.py new file mode 100644 index 000000000..0ca456a78 --- /dev/null +++ b/extras/volfilter.py @@ -0,0 +1,167 @@ +# Copyright (c) 2010-2011 Red Hat, Inc. +# +# This file is part of HekaFS. +# +# HekaFS is free software: you can redistribute it and/or modify it under the +# terms of the GNU General Public License, version 3, as published by the Free +# Software Foundation. +# +# HekaFS is distributed in the hope that it will be useful, but WITHOUT ANY +# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR +# A PARTICULAR PURPOSE. See the GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License * along +# with HekaFS. If not, see <http://www.gnu.org/licenses/>. + +import copy +import string +import sys +import types + +good_xlators = [ + "cluster/afr", + "cluster/dht", + "cluster/distribute", + "cluster/replicate", + "cluster/stripe", + "debug/io-stats", + "features/access-control", + "features/locks", + "features/marker", + "features/uidmap", + "performance/io-threads", + "protocol/client", + "protocol/server", + "storage/posix", +] + +def copy_stack (old_xl,suffix,recursive=False): + if recursive: + new_name = old_xl.name + "-" + suffix + else: + new_name = suffix + new_xl = Translator(new_name) + new_xl.type = old_xl.type + # The results with normal assignment here are . . . amusing. + new_xl.opts = copy.deepcopy(old_xl.opts) + for sv in old_xl.subvols: + new_xl.subvols.append(copy_stack(sv,suffix,True)) + # Patch up the path at the bottom. + if new_xl.type == "storage/posix": + new_xl.opts["directory"] += ("/" + suffix) + return new_xl + +def cleanup (parent, graph): + if parent.type in good_xlators: + # Temporary fix so that HekaFS volumes can use the + # SSL-enabled multi-threaded socket transport. + if parent.type == "protocol/server": + parent.type = "protocol/server2" + parent.opts["transport-type"] = "ssl" + elif parent.type == "protocol/client": + parent.type = "protocol/client2" + parent.opts["transport-type"] = "ssl" + sv = [] + for child in parent.subvols: + sv.append(cleanup(child,graph)) + parent.subvols = sv + else: + parent = cleanup(parent.subvols[0],graph) + return parent + +class Translator: + def __init__ (self, name): + self.name = name + self.type = "" + self.opts = {} + self.subvols = [] + self.dumped = False + def __repr__ (self): + return "<Translator %s>" % self.name + +def load (path): + # If it's a string, open it; otherwise, assume it's already a + # file-like object (most notably from urllib*). + if type(path) in types.StringTypes: + fp = file(path,"r") + else: + fp = path + all_xlators = {} + xlator = None + last_xlator = None + while True: + text = fp.readline() + if text == "": + break + text = text.split() + if not len(text): + continue + if text[0] == "volume": + if xlator: + raise RuntimeError, "nested volume definition" + xlator = Translator(text[1]) + continue + if not xlator: + raise RuntimeError, "text outside volume definition" + if text[0] == "type": + xlator.type = text[1] + continue + if text[0] == "option": + xlator.opts[text[1]] = string.join(text[2:]) + continue + if text[0] == "subvolumes": + for sv in text[1:]: + xlator.subvols.append(all_xlators[sv]) + continue + if text[0] == "end-volume": + all_xlators[xlator.name] = xlator + last_xlator = xlator + xlator = None + continue + raise RuntimeError, "unrecognized keyword %s" % text[0] + if xlator: + raise RuntimeError, "unclosed volume definition" + return all_xlators, last_xlator + +def generate (graph, last, stream=sys.stdout): + for sv in last.subvols: + if not sv.dumped: + generate(graph,sv,stream) + print >> stream, "" + sv.dumped = True + print >> stream, "volume %s" % last.name + print >> stream, " type %s" % last.type + for k, v in last.opts.iteritems(): + print >> stream, " option %s %s" % (k, v) + if last.subvols: + print >> stream, " subvolumes %s" % string.join( + [ sv.name for sv in last.subvols ]) + print >> stream, "end-volume" + +def push_filter (graph, old_xl, filt_type, opts={}): + suffix = "-" + old_xl.type.split("/")[1] + if len(old_xl.name) > len(suffix): + if old_xl.name[-len(suffix):] == suffix: + old_xl.name = old_xl.name[:-len(suffix)] + new_xl = Translator(old_xl.name+suffix) + new_xl.type = old_xl.type + new_xl.opts = old_xl.opts + new_xl.subvols = old_xl.subvols + graph[new_xl.name] = new_xl + old_xl.name += ("-" + filt_type.split("/")[1]) + old_xl.type = filt_type + old_xl.opts = opts + old_xl.subvols = [new_xl] + graph[old_xl.name] = old_xl + +def delete (graph, victim): + if len(victim.subvols) != 1: + raise RuntimeError, "attempt to delete non-unary translator" + for xl in graph.itervalues(): + while xl.subvols.count(victim): + i = xl.subvols.index(victim) + xl.subvols[i] = victim.subvols[0] + +if __name__ == "__main__": + graph, last = load(sys.argv[1]) + generate(graph,last) diff --git a/extras/who-wrote-glusterfs/gitdm.aliases b/extras/who-wrote-glusterfs/gitdm.aliases new file mode 100644 index 000000000..784a3e3bc --- /dev/null +++ b/extras/who-wrote-glusterfs/gitdm.aliases @@ -0,0 +1,48 @@ +# +# This is the email aliases file, mapping secondary addresses onto a single, +# canonical address. This file should probably match the contents of .mailmap +# in the root of the git repository. +# +# Format: <alias> <real> + +amar@gluster.com amarts@redhat.com +amar@del.gluster.com amarts@redhat.com +avati@amp.gluster.com avati@redhat.com +avati@blackhole.gluster.com avati@redhat.com +avati@dev.gluster.com avati@redhat.com +avati@gluster.com avati@redhat.com +wheelear@gmail.com awheeler@redhat.com +anush@gluster.com ashetty@redhat.com +csaba@gluster.com csaba@redhat.com +csaba@lowlife.hu csaba@redhat.com +csaba@zresearch.com csaba@redhat.com +harsha@gluster.com fharshav@redhat.com +harsha@zresearch.com fharshav@redhat.com +harsha@dev.gluster.com fharshav@redhat.com +harsha@harshavardhana.net fharshav@redhat.com +kkeithle@f16node1.kkeithle.usersys.redhat.com kkeithle@redhat.com +kaushal@gluster.com kaushal@redhat.com +kaushikbv@gluster.com kbudiger@redhat.com +krishna@gluster.com ksriniva@redhat.com +krishna@zresearch.com ksriniva@redhat.com +krishna@guest-laptop ksriniva@redhat.com +kp@gluster.com kparthas@redhat.com +me@louiszuckerman.com louiszuckerman@gmail.com +msvbhat@gmail.com vbhat@redhat.com +vishwanath@gluster.com vbhat@redhat.com +pavan@dev.gluster.com pavan@gluster.com +zaitcev@yahoo.com zaitcev@kotori.zaitcev.us +pranithk@gluster.com pkarampu@redhat.com +raghavendrabhat@gluster.com raghavendra@redhat.com +raghavendra@gluster.com rgowdapp@redhat.com +raghavendra@zresearch.com rgowdapp@redhat.com +rahulcssjce@gmail.com rahulcs@redhat.com +rajesh@gluster.com rajesh@redhat.com +rajesh.amaravathi@gmail.com rajesh@redhat.com +shehjart@zresearch.com shehjart@gluster.com +venky@gluster.com vshankar@redhat.com +vijay@gluster.com vbellur@redhat.com +vijay@dev.gluster.com vbellur@redhat.com +vijaykumar.koppad@gmail.com vkoppad@redhat.com +vikas@zresearch.com vikas@gluster.com +shishirng@gluster.com sgowda@redhat.com diff --git a/extras/who-wrote-glusterfs/gitdm.config b/extras/who-wrote-glusterfs/gitdm.config new file mode 100644 index 000000000..e1ff2bd5b --- /dev/null +++ b/extras/who-wrote-glusterfs/gitdm.config @@ -0,0 +1,8 @@ +# +# This is the gitdm configuration file for GlusterFS. +# See the gitdm.config in the gitdm repositofy for additional options and +# comments. +# + +EmailAliases gitdm.aliases +EmailMap gitdm.domain-map diff --git a/extras/who-wrote-glusterfs/gitdm.domain-map b/extras/who-wrote-glusterfs/gitdm.domain-map new file mode 100644 index 000000000..f1c305898 --- /dev/null +++ b/extras/who-wrote-glusterfs/gitdm.domain-map @@ -0,0 +1,15 @@ +# +# Here is a set of mappings of domain names onto employer names. +# +active.by ActiveCloud +cern.ch CERN +gluster.com Red Hat +gooddata.com GoodData +hastexo.com hastexo +ibm.com IBM +linbit.com LINBIT +netbsd.org NetBSD +netdirect.ca Net Direct +redhat.com Red Hat +stepping-stone.ch stepping stone GmbH +zresearch.com Red Hat diff --git a/extras/who-wrote-glusterfs/who-wrote-glusterfs.sh b/extras/who-wrote-glusterfs/who-wrote-glusterfs.sh new file mode 100755 index 000000000..487f5874b --- /dev/null +++ b/extras/who-wrote-glusterfs/who-wrote-glusterfs.sh @@ -0,0 +1,50 @@ +#!/bin/sh +# +# Gather statistics on "Who wrote GlusterFS". The idea comes from the excellent +# articles on http://lwn.net/ named "Who wrote <linux-version>?". +# +# gitdm comes from git://git.lwn.net/gitdm.git by Jonathan Corbet. +# +# Confguration files used: +# - gitdm.config: main configuration file, pointing to the others +# - gitdm.aliases: merge users with different emailaddresses into one +# - gitdm.domain-map: map domain names from emailaddresses to companies +# + +DIRNAME=$(dirname $0) + +GITDM_REPO=git://git.lwn.net/gitdm.git +GITDM_DIR=${DIRNAME}/gitdm +GITDM_CMD="python ${GITDM_DIR}/gitdm" + +error() +{ + local ret=${?} + printf "${@}\n" > /dev/stderr + return ${ret} +} + +check_gitdm() +{ + if [ ! -e "${GITDM_DIR}/gitdm" ] + then + git clone --quiet git://git.lwn.net/gitdm.git ${DIRNAME}/gitdm + fi +} + +# The first argument is the revision-range (see 'git rev-list --help'). +# REV can be empty, and the statistics will be calculated over the whole +# current branch. +REV=${1} +shift +# all remaining options are passed to gitdm, see the gitdm script for an +# explanation of the accepted options. +GITDM_OPTS=${@} + +if ! check_gitdm +then + error "Could not find 'gitdm', exiting..." + exit 1 +fi + +git log --numstat -M ${REV} | ${GITDM_CMD} -b ${DIRNAME} -n ${GITDM_OPTS} |