glusterfs.git/rpc/rpc-transport/socket/src, branch v3.7.4

SSL improvements: do not fail if certificate purpose is set

2015-08-24T06:19:45+00:00

Since glusterfs shares the same settings for client-side
and server-side of SSL, we need to ignore any certificate
usage specification (SSL client vs SSL server), otherwise
SSL connexions will fail with 'unsupported cerritifcate"

Backport of I7ef60271718d2d894176515aa530ff106127bceb

BUG: 1247153
Change-Id: I04e2f50dafd84d6eee15010f045016c91a0e1aac
Signed-off-by: Emmanuel Dreyfus 
Reviewed-on: http://review.gluster.org/11842
Tested-by: Gluster Build System 
Tested-by: NetBSD Build System 
Reviewed-by: Kaleb KEITHLEY 
Reviewed-by: Jeff Darcy

SSL improvements: ECDH, DH, CRL, and accessible options

2015-08-05T11:52:04+00:00

- Introduce ssl.dh-param option to specify a file containinf DH parameters.
  If it is provided, EDH ciphers are available.

- Introduce ssl.ec-curve option to specify an elliptic curve name. If
  unspecified, ECDH ciphers are available using the prime256v1 curve.

- Introduce ssl.crl-path option to specify the directory where the
  CRL hash file can be found. Setting to NULL disable CRL checking,
  just like the default.

- Make all ssl.* options accessible through gluster volume set.

- In default cipher list, exclude weak ciphers instead of listing
  the strong ones.

- Enforce server cipher preference.

- introduce RPC_SET_OPT macro to factor repetitive code in glusterd-volgen.c

- Add ssl-ciphers.t test to check all the features touched by this change.

Backport of I7bfd433df6bbf176f4a58e770e06bcdbe22a101a

Change-Id: I2947eabe76ae0487ecad52a60befb7de473fc90c
BUG: 1247153
Signed-off-by: Emmanuel Dreyfus @
Reviewed-on: http://review.gluster.org/11763
Tested-by: NetBSD Build System 
Reviewed-by: Jeff Darcy

rpc: fix binding brick issue while bind-insecure is enabled

2015-07-27T04:31:56+00:00

This patch is backport of http://review.gluster.org/#/c/11512/

> problem:

> When bind-insecure is turned on (which is the default now), it may happen
> that brick is not able to bind to port assigned by Glusterd for example
> 49192-49195...
>
> It seems to occur because the rpc_clnt connections are binding to ports in
> the same range. so brick fails to bind to a port which is already used by
> someone else
>
> solution:
>
> fix for now is to  make rpc_clnt to get port numbers from 65535 in a
> descending
> order, as a result port clash is minimized
>
> other fixes:
>
> previously rdma binds to port >= 1024 if it cannot find a free port < 1024,
> even when bind insecure was turned off(ref to commit '0e3fd04e'), this patch
> add's a check for bind-insecure in gf_rdma_client_bind function
>
> This patch also re-enable bind-insecure and allow insecure by default
> which was reverted (ref: commit cef1720) previously

> Change-Id: Ia1cfa93c5454e2ae0ff57813689b75de282ebd07
> BUG: 1238661
> Signed-off-by: Prasanna Kumar Kalever 

Change-Id: Iea55f9b2a57b5e24d3df2c5fafae12fe99e9dee0
BUG: 1246481
Signed-off-by: Prasanna Kumar Kalever 
Reviewed-on: http://review.gluster.org/11758
Tested-by: NetBSD Build System 
Tested-by: Gluster Build System 
Reviewed-by: Raghavendra G

rpc,server,glusterd: Init transport list for accepted transport

2015-07-26T17:36:39+00:00

GlusterD or a brick would crash when encrypted transport was enabled and
an unencrypted client tried to connect to them. The crash occured when
GlusterD/server tried to remove the transport from their xprt_list due
to a DISCONNECT event. But as the client transport's list head wasn't
inited, the process would crash when list_del was performed.

Initing the client transports list head during acceptence, prevents this
crash.

Also, an extra check has been added to the GlusterD and Server
notification handlers for client DISCONNECT events. The handlers will
now first check if the client transport is a member of any list.
GlusterD and Server DISCONNECT event handlers could be called without
the ACCEPT handler, which adds the transport to the list, being called.
This situation also occurs when an unencrypted client tries to establish
a connection with an encrypted server.

Change-Id: Icc24a08d60e978aaa1d3322e0cbed680dcbda2b4
BUG: 1246809
Signed-off-by: Kaushal M 
Reviewed-on: http://review.gluster.org/11692
Tested-by: Gluster Build System 
Tested-by: NetBSD Build System 
Reviewed-by: Raghavendra G 
(cherry picked from commit a909ccfa1b4cbf656c4608ef2124347851c492cb)
Reviewed-on: http://review.gluster.org/11762

rpc-transport: socket_poller fixes for proper working of mgmt encryption

2015-07-14T10:25:42+00:00

  Backport of 8c39f14 from master

socket_poller, the polling function used by ssl own_thread, had two
issues which lead to GlusterD crashes when using management encryption

Issue 1
-------
socket_poller calls functions which require THIS to be set. But, THIS
was being set conditionally. Because of this, functions could sometimes
be called without THIS being set. For example, rpc_transport_notify
could be called for an accepted client socket without THIS being set, as
THIS was only set it the transport wasn't yet connected. This would
cause the process to crash when THIS was accessed by the called
functions.

To fix this, THIS is being set at the start of socket_poller
unconditionally.

Issue 2
-------
DISCONNECT notify was being sent on the listener transport instead of
the client transport. The DISCONNECT event was converted to a
LISTENER_DEAD event in rpcsvc_handle_disconnect, as it could not find
the listener socket of the listener socket. GlusterD was notified of a
LISTENER_DEAD event instead of a DISCONNECT and failed to remove the
client transport from its xprt_list. The transport would subsequently
be freed, leaving the xprt_list with a corrupted/invalid entry. Later,
when GlusterD would iterate over the xprt_list to send notifications, it
would crash when the invalid entry was accessed.

To fix this, DISCONNECT notification in socket_poller is sent on the
client socket, as it is done in the epoll handler.

Change-Id: I0370b7c6d7eb13de10ebf08d91a4a39dc7d64c7a
BUG: 1242734
Signed-off-by: Kaushal M 
Reviewed-on: http://review.gluster.org/11655
Tested-by: NetBSD Build System 
Tested-by: Gluster Build System 
Reviewed-by: Krishnan Parthasarathi

Revert "rpc: By default set allow-insecure, bind-insecure to on"

2015-07-02T15:46:13+00:00

This reverts commit 243a5b429f225acb8e7132264fe0a0835ff013d5.

This patch introduced a regression where client no longer binds to 
privileged port. This is causing lots of regressions. Hence reverting 
this patch for now and will be resent after suitable modifications.

Change-Id: I302252fd3832b0a5a03b04e30cfa0def37597404
Reviewed-on: http://review.gluster.org/11508
Reviewed-by: Raghavendra G 
Tested-by: Raghavendra G

rpc: By default set allow-insecure, bind-insecure to on

2015-07-01T06:59:51+00:00

        Backport of http://review.gluster.org/11039

since we now use SSL (Secure Sockets Layer) for the security issues, the patch
changes the default setting to allow connections/requests from non-privilaged
ports by setting allow-insecure and bind-insecure to 1

Also added bind functionality for insecure binding which can select from
available local ports dynamically

BUG: 1232660
Change-Id: I927e112223f33611452093e38cd846a0b9347e57
Signed-off-by: Pranith Kumar K 
Signed-off-by: Prasanna Kumar Kalever 
Reviewed-on: http://review.gluster.org/11274
Tested-by: NetBSD Build System 
Tested-by: Gluster Build System 
Reviewed-by: Raghavendra G

build: fix compiling on older distributions

2015-06-11T14:29:03+00:00

data-tiering is disabled on RHEL-5 because it depends on a too new
SQLite version.

This change also prevents installing some of files that are used by
geo-replication, which is also not available on RHEL-5. geo-replication
depends on a too recent version of Python.

Due to an older version of OpenSSL, some of the newer functions can not
be used. A fallback to previous functions is done. Unfortunately RHEL-5
does not seem to have TLSv1.2 support, so only older versions can be
used.

Cherry picked from commit 0209b18fd65f9df5ebd0a8764ebf864d0d392998:
> Change-Id: I672264a673f5432358d2e83b17e2a34efd9fd913
> BUG: 1222317
> Signed-off-by: Niels de Vos 
> Reviewed-on: http://review.gluster.org/10803
> Tested-by: NetBSD Build System 
> Tested-by: Gluster Build System 
> Reviewed-by: Kaleb KEITHLEY 
> Reviewed-by: Vijay Bellur 

Also including the changes from http://review.gluster.org/11140:

> build: improve detection of new OpenSSL features
>
> Building on Mac OS X revealed that the current check for
> CRYPTO_THREADID_set_callback() availability in OpenSSL is not correct.
>
> There also does not seem to be a guarantee that TLSv1_2_method() is
> available when TLS1_2_VERSION is #define'd.
>
> Change-Id: I21508065fc181a1c74bee4fd6d23bb5bdf7cea7a
> BUG: 1222317
> Reviewed-on: http://review.gluster.org/11140
> Original-author: Kaleb KEITHLEY 
> Signed-off-by: Niels de Vos 

Change-Id: I672264a673f5432358d2e83b17e2a34efd9fd913
BUG: 1228510
Signed-off-by: Niels de Vos 
Reviewed-on: http://review.gluster.org/11096
Tested-by: Gluster Build System 
Reviewed-by: Atin Mukherjee 
Reviewed-by: Kaleb KEITHLEY

socket: throttle only connected transport

2015-06-04T11:22:02+00:00

BUG: 1227167
Change-Id: I2352c6661ae5aee8cf06e572ec0462b76901d46f
Signed-off-by: Krishnan Parthasarathi 
Reviewed-on: http://review.gluster.org/10887
Reviewed-by: Raghavendra G 
Reviewed-on: http://review.gluster.org/11038
Reviewed-by: Shyamsundar Ranganathan 
Tested-by: Gluster Build System 
Tested-by: Raghavendra G

socket: use OpenSSL multi-threading interfaces

2015-04-09T09:55:37+00:00

OpenSSL isn't thread-safe unless you register these locking and thread
ID functions.  Most often the crashes would occur around
X509_verify_cert, even though it's insane that the certificate parsing
functions wouldn't be thread-safe.  The bug for this was filed over
two years ago, but it didn't seem like a high priority because the bug
didn't bite anyone until it caused a spurious regression-test failure.
Ironically, that was on a test for a *different* spurious
regression-test failure, which I guess is just deserts[1] for leaving
this on the to-do list so long.

[1] Yes, it really is "deserts" in that phrase - not as in very dry
places, but from late Latin "deservire" meaning to serve well or
zealously.  Aren't commit messages educational?

Change-Id: I2a6c0e9b361abf54efa10ffbbbe071404f82b0d9
BUG: 906763
Signed-off-by: Jeff Darcy 
Reviewed-on: http://review.gluster.org/10075
Tested-by: Gluster Build System 
Reviewed-by: Kaleb KEITHLEY 
Reviewed-by: Vijay Bellur