glusterfs.git, branch v3.6.3beta1

glusterd: Fix spurious volume delete failure

2015-02-11T11:45:36+00:00

If volume uses quota, volume delete operation should unmount the
auxiliary quota mount usin glusterd_remove_auxiliary_mount(). This
may fail with EBADF is the mount is already gone. In that situation,
ignore the error so that volume delete succeeds.

This fixes a spurious failure on NetBSD in tests/basic/quota.t 74-75

Backport of I69325f71fc2c8af254db46f696c8669a4e6bd7e4

BUG: 1138897
Change-Id: If0d382d44a956bb9fd8c41299f82affdf2ee0618
Signed-off-by: Emmanuel Dreyfus 
Reviewed-on: http://review.gluster.org/9484
Tested-by: Gluster Build System 
Reviewed-by: Krishnan Parthasarathi 
Tested-by: Krishnan Parthasarathi

storage/posix: Don't try to set gfid in case of INTERNAL-mknod

2015-02-11T09:49:16+00:00

        Backport of http://review.gluster.org/9446

BUG: 1184527
Change-Id: I3e31b0672704cae09c5097f84229b8264a6c0fbe
Signed-off-by: Pranith Kumar K 
Reviewed-on: http://review.gluster.org/9472
Tested-by: Gluster Build System 
Reviewed-by: Raghavendra Bhat

storage/posix: Set gfid after all xattrs, uid/gid are set

2015-02-11T09:49:02+00:00

        Backport of http://review.gluster.com/9434

Problem:
When a new entry is created gfid is set even before uid/gid, xattrs
are set on the entry. This can lead to dht/afr healing that file/dir
with the uid/gid it sees just after the gfid is set, i.e. root/root.
Sometimes setattr/setxattr are failing on that file/dir.

Fix:
Set gfid of the file/directory only after uid/gid, xattrs are setup
properly. Readdirp, lookup either wait for the gfid to be assigned
to the entry or not update the in-memory inode ctx in posix-acl
xlator which was producing lot EACCESS/EPERM to the application
or dht/afr self-heals.

BUG: 1184527
Change-Id: Ia6dfd492e03db2275665e7f63260611b310e38e6
Signed-off-by: Pranith Kumar K 
Reviewed-on: http://review.gluster.org/9471
Tested-by: Gluster Build System 
Reviewed-by: Raghavendra Bhat

ec: Don't use inodelk on getxattr when clearing locks

2015-02-11T09:40:11+00:00

When command 'clear-locks' from cli is executed, a getxattr request
is received by ec. This request was handled as usual, first locking
the inode. Once this request was processed by the bricks, all locks
were removed, including the lock used by ec.

When ec tried to unlock the previously acquired lock (which was
already released), caused a crash in glusterfsd.

This fix executes the getxattr request without any lock acquired
for the clear-locks command.

This is a backport of http://review.gluster.org/9440/

Change-Id: I77e550d13c4673d2468a1e13fe6e2fed20e233c6
BUG: 1181977
Signed-off-by: Xavier Hernandez 
Reviewed-on: http://review.gluster.org/9444
Reviewed-by: Dan Lambright 
Tested-by: Gluster Build System 
Reviewed-by: Raghavendra Bhat

ec: Fix posix compliance failures

2015-02-11T09:37:55+00:00

This patch solves some problems that caused dispersed volumes to not
pass posix smoke tests:

* Problems in open/create with O_WRONLY
    Opening files with -w- permissions using O_WRONLY returned an EACCES
    error because internally O_WRONLY was replaced with O_RDWR.

* Problems with entrylk on renames.
    When source and destination were the same, ec tried to acquire
    the same entrylk twice, causing a deadlock.

* Overwrite of a variable when reordering locks.
    On a rename, if the second lock needed to be placed at the beggining
    of the list, the 'lock' variable was overwritten and later its timer
    was cancelled, cancelling the incorrect one.

* Handle O_TRUNC in open.
    When O_TRUNC was received in an open call, it was blindly propagated
    to child subvolumes. This caused a discrepancy between real file
    size and the size stored into trusted.ec.size xattr. This has been
    solved by removing O_TRUNC from open and later calling ftruncate.

This is a backport of http://review.gluster.org/9420

Change-Id: I20c3d6e1c11be314be86879be54b728e01013798
BUG: 1159471
Signed-off-by: Xavier Hernandez 
Reviewed-on: http://review.gluster.org/9501
Tested-by: Gluster Build System 
Reviewed-by: Dan Lambright 
Reviewed-by: Pranith Kumar Karampuri 
Reviewed-by: Raghavendra Bhat

afr : Fixes to 59ba78ae1461651e290ce72013786d828545d4c1

2015-02-11T09:35:38+00:00

A bug was found while reviewing
http://review.gluster.org/9377/
(inode was not being unreffed after use).
It is fixed on master as a part of the
following change - 
http://review.gluster.org/#/c/9377/7/xlators/cluster/afr/src/afr-common.c .
Fixing the same issue in 3.6 with this patch.

Change-Id: Ibdc4be49d88613ccb1f80349eb4d368710c0c24b
BUG: 1173528
Signed-off-by: Anuradha 
Reviewed-on: http://review.gluster.org/9450
Reviewed-by: Niels de Vos 
Tested-by: Gluster Build System 
Reviewed-by: Ravishankar N 
Reviewed-by: Raghavendra Bhat

cluster/afr: When parent and entry read subvols are different, set entry->inode to NULL

2015-02-11T09:30:02+00:00

        Backport of: http://review.gluster.org/#/c/9477

That way a lookup would be forced on the entry, and its attributes will
always be selected from its read subvol.

Change-Id: Iaba25e2cd5f83e983fc8b1a1f48da3850808e6b8
BUG: 1186119
Signed-off-by: Krutika Dhananjay 
Reviewed-on: http://review.gluster.org/9562
Reviewed-by: Pranith Kumar Karampuri 
Tested-by: Gluster Build System 
Reviewed-by: Raghavendra Bhat

ec: Fix failures with missing files

2015-02-11T09:20:24+00:00

When a file does not exist on a brick but it does on others, there
could be problems trying to access it because there was some loc_t
structures with null 'pargfid' but 'name' was set. This forced
inode resolution based on /name instead of  which
would be the correct one. To solve this problem, 'name' is always
set to NULL when 'pargfid' is not present.

Another problem was caused by an incorrect management of errors
while doing incremental locking. The only allowed error during an
incremental locking was ENOTCONN, but missing files on a brick can
be returned as ESTALE. This caused an EIO on the operation.

This patch doesn't care of errors during an incremental locking. At
the end of the operation it will check if there are enough successfully
locked bricks to continue or not.

This is a backport of http://review.gluster.org/9407/

Change-Id: I9360ebf8d819d219cea2d173c09bd37679a6f15a
BUG: 1183716
Signed-off-by: Xavier Hernandez 
Reviewed-on: http://review.gluster.org/9560
Tested-by: Gluster Build System 
Reviewed-by: Dan Lambright 
Reviewed-by: Raghavendra Bhat

NetBSD portability fix: recover errno on runner error

2015-02-04T12:06:45+00:00

Some time ago we introduced F_CLOSEM to efficiently close unused
file descriptors when using a runner. But since it also close the
file descriptor used to send back errno to glusterd, it got unable
to detect an error on execve().

Fix this by backing out F_CLOSEM usage.

BUG: 1138897
Change-Id: I147a0924e8b1721c2a5ff67c3b943579d3a03bf5
Signed-off-by: Emmanuel Dreyfus 
Reviewed-on: http://review.gluster.org/9448
Tested-by: Gluster Build System 
Reviewed-by: Niels de Vos 
Reviewed-by: Vijay Bellur 
Reviewed-by: Raghavendra Bhat

cluster/ec: Do not modify quota, selinux xattrs in healing

2015-02-04T11:54:47+00:00

        Backport of http://review.gluster.org/9401

Problem:
EC heal tries to heal quota-size, selinux xattrs as well.  quota-size is
private to the brick but since quotad accesses them using the standard
interface as well, they can not be filtered in the fops.

Fix:
Ignore QUOTA_SIZE_KEY and SELINUX xattrs during heal.

BUG: 1178590
Change-Id: Id569a49ef996e5507f4474c99b6cdc22781ad82d
Signed-off-by: Pranith Kumar K 
Reviewed-on: http://review.gluster.org/9454
Reviewed-by: Xavier Hernandez 
Tested-by: Gluster Build System 
Reviewed-by: Raghavendra Bhat