glusterfs.git/xlators/mgmt/glusterd, branch release-3.2

mgmt/glusterd: check brickinfo for NULL before dereferencing it

2013-01-08T05:58:56+00:00

Change-Id: I59fed6adf13d6d72d79c986bf3578ca79b9756ac
BUG: 781953
Signed-off-by: Raghavendra Bhat 
Reviewed-on: http://review.gluster.org/2648
Tested-by: Gluster Build System 
Reviewed-by: Krishnan Parthasarathi

Most commands cause glusterd 3.2.7 to crash on Debian Wheezy

2012-07-20T23:13:10+00:00

When xlators/mgmt/glusterd/src/glusterd-rpc-ops.c is compiled with
-O2 by gcc-4.7.1, the compiler optimizes away all the rsp.foo = bar
statements because rsp goes out of scope at the end of each of
the blocks in the switch statement in glusterd-op_send_cli_response().

That is:
    void *cli_rsp = NULL;

    switch(foo) {
    case BAR:
    {
        struct bar rsp = {0,};
        /* init statements here are optimized away */
        rsp.x = 0;
        rsp.y = 1;
        cli_rsp = &rsp;
    }
    ...
    }
    /* don't expect cli_rsp to point at anything meaningful here */

This particular idiom hasn't been a problem thus far, e.g. even with
gcc-4.7.0 in Fedora17 also compiling with -O2. That not withstanding,
using this idiom has probably always been on shaky ground; semantically
it is correct that rsp goes out of scope at the end of the block.

Note: I have not surveyed the source to see whether this idiom appears
anywhere else.

Note also that glusterd-op_send_cli_response() has been rewritten around
the refactored gf_cli_rsp for 3.3.x and later and this particular bug is
not in 3.3.x and later releases.

BUG: 837684
Change-Id: I82f61ad2b9827c5b96af14b180a82c3ab350f559
Signed-off-by: Kaleb S. KEITHLEY 
Reviewed-on: http://review.gluster.com/3707
Tested-by: Gluster Build System 
Reviewed-by: Jeff Darcy 
Reviewed-by: Anand Avati

Fixes bug-765574, failure to build from hardened source

2012-01-26T10:19:30+00:00

Glusterfs 3.2.5 failed to build on Debian which prevented the
Debian project from updating their package to the latest release
version.

There were two changes needed, in both cases just adding a
printf format string argument to a call to gf_log.

The compiler error was that the argument was neither a string
literal nor a format string.  It was instead a pointer to a string.
The obvious solution was to add a format argument of "%s" before
the string pointer argument.

This has been tested successfully on Debian Sid.

Change-Id: I62c8a46938184c8a37224c2481bd1de00369ab19
Signed-off-by: louis zuckerman 
Reviewed-on: http://review.gluster.com/2598
Tested-by: Gluster Build System 
Reviewed-by: Kaleb KEITHLEY 
Reviewed-by: Amar Tumballi

mgmt/glusterd: do not assert if op_ctx is NULL

2012-01-24T16:37:03+00:00

In glusterd_op_send_cli_response we are asserting if op_ctx is NULL.
But in handler functions of different ops, we are sending op_ctx NULL
which is valid, but still glusterd will assert.

Change-Id: I37824c694d14e87a234389d639dfb000592f9513
BUG: 783913
Signed-off-by: Raghavendra Bhat 
Reviewed-on: http://review.gluster.com/2678
Tested-by: Gluster Build System 
Reviewed-by: Krishnan Parthasarathi 
Reviewed-by: Anand Avati

cli: add geo-replication log-rotate command

2011-11-04T09:06:44+00:00

Rotating geo-replication master/monitor log files from cli.
On invocation, the log file for a given master-slave session
is backed up with the current timestamp suffixed to the file
name and signal is sent to gsyncd to start logging to a new
log file.

Sample commands:

* Rotate log file for this : session:
gluster volume geo-replication   log-rotate

* Rotate log files for all session for master volume 
gluster volume geo-replication  log-rotate

* Rotate log files for all sessions:
gluster volume geo-replication log-rotate

Change-Id: If801743e9f37bd282e68d262203141626ce77e55
BUG: 3519
Reviewed-on: http://review.gluster.com/653
Tested-by: Gluster Build System 
Reviewed-by: Csaba Henk

glusterd rebalance: fail rebalance command on non-distribute volumes

2011-11-03T13:15:18+00:00

Change-Id: Id8cf3cbc8297b5f24f67c152d789b7c2ac299ccc
BUG: 3729
Reviewed-on: http://review.gluster.com/652
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

glusterd: Fixed volume profile's "are all bricks down?" algo.

2011-10-28T15:08:13+00:00

"count" key is not set if no brick in the volume (located across peers) is
running. Sending "count" even when zero bricks are running from originator
glusterd to client, simplifies "are all bricks down?" algo.

Change-Id: I3a8db48c4c61cd31ba96031638a3c3d26efd2887
BUG: 3553
Reviewed-on: http://review.gluster.com/641
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

glusterd: Removing delayed moving of op sm for stop vol/remove brick op.

2011-10-28T11:58:18+00:00

Earlier we would wait until brick disconnect or 5s whichever was earlier,
before we move op sm from brick op stage to commit stage. This involves a race
where both the above mentioned events can happen 'concurrently' and result in
double free of the event context.

Change-Id: Ia369994448e71a052679d2373b8ab9177d56a541
BUG: 3700
Reviewed-on: http://review.gluster.com/626
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

glusterd: Fail volume 'modify' operations when rb is ongoing.

2011-10-24T13:17:34+00:00

add-brick, stop-volume, remove-brick are the operations that are explicitly
'failed' when attempted while replace-brick is in progress.

Change-Id: Ia8034177e03a9ab10f9e8d5c8bc417427a1218e3
BUG: 3738
Reviewed-on: http://review.gluster.com/631
Tested-by: Gluster Build System 
Reviewed-by: Amar Tumballi 
Reviewed-by: Vijay Bellur

glusterd: Refactored rb subcmds code and fixed some minor issues.

2011-10-19T10:19:56+00:00

Change-Id: Ie14492451cab821e7ed60e68dbaff22d7d78fba9
BUG: 3709
Reviewed-on: http://review.gluster.com/609
Tested-by: Gluster Build System 
Reviewed-by: Amar Tumballi 
Reviewed-by: Vijay Bellur