glusterfs.git/geo-replication/syncdaemon, branch v3.7.6

geo-rep: Update geo-rep status, if monitor process is killed

2015-11-01T14:57:08+00:00

Problem:
When the monitor process itself is getting killed, geo-rep session
still shows as active.

Status command will just pick up the content from the status file
to show the output. Monitor process is the one which updates the Status file.

When the monitor process itself gets killed, there is no way to update
the status file. So, geo-rep session status command ends up showing
last updated Status present in the status file. 

Solution:
While getting the status output, check whether monitor process is running.
If it is NOT running, update the status as STOPPED.

Change-Id: I86a7ac1746dd8f27eef93658e992ef16f6068d9d
BUG: 1276060
Signed-off-by: Saravanakumar Arumugam 
Reviewed-on: http://review.gluster.org/11873
Tested-by: NetBSD Build System 
Tested-by: Gluster Build System 
Reviewed-by: Milind Changire 
Reviewed-by: Kotresh HR 
Reviewed-by: Jeff Darcy 
(cherry picked from commit 4d4c7d5dc54850dcf916083b2b1398d9bfe2bfe6)
Reviewed-on: http://review.gluster.org/12448
Reviewed-by: Aravinda VK 
Reviewed-by: Vijay Bellur

geo-rep: Add data operation if mknod with tier attribute

2015-10-29T07:18:27+00:00

This is a series of patches which aims to fix geo-replication
in a Tiering Volume.

Problem:
Consider, a file is placed in volume initially and then hot tier is
attached. During any operation on the file, due to lookup a linkto
file is created in hot tier.

Now, any namespace operation carried out on the file is recorded in
both cold and hot tier.
There is a room for races when both changelogs are replayed.

Solution:
So, We are going to replay (namespace related)operations
only in the hot tier.

Why?
a. If the file is directly placed in Hot tier, all fops will be
recorded in HOT tier.

b. If  the file is already present in Cold tier, and if any fop is
carried out, it creates linkto file in Hot tier.
Now, operations like UNLINK, RENAME are captured in Hot tier(by means of linkto file).

This way, we can get both tier's operation in HOT tier itself.

But, We may miss initial Data sync immediately after creating the
file as it is only recording MKNOD. So, if MKNOD encountered
with sticky bit set, queue DATA operation for the corresponding gfid.
(This is addressed here in this patch)

So, If tier-gfid linkto is set, we need to record the corresponding
MKNOD. Earlier this was avoided as it was set as INTERNAL fop.
(This changelog related changes are addressed in the patch:
 - http://review.gluster.org/12417)

Change-Id: I2fa84cfa2b0f86506c3d15d484138ab9651e4f83
BUG: 1275173
Signed-off-by: Saravanakumar Arumugam 
Reviewed-on: http://review.gluster.org/12326
Tested-by: NetBSD Build System 
Tested-by: Gluster Build System 
Reviewed-by: Kotresh HR 
Reviewed-by: Aravinda VK 
(cherry picked from commit ffc39c9d8807464b5c78959bc43dc12b22f5a37b)
Reviewed-on: http://review.gluster.org/12427
Reviewed-by: Venky Shankar

geo-rep: Do not crash worker on ESTALE

2015-08-06T06:17:25+00:00

Handle ESTALE returned by lstat gracefully
by retrying it. Do not crash the worker.

BUG: 1249547
Change-Id: I57fb9933900153ab41c3d9b73748b1cdaa8d89ca
Reviewed-on: http://review.gluster.org/11772
Tested-by: Gluster Build System 
Tested-by: NetBSD Build System 
Reviewed-by: Aravinda VK 
Reviewed-by: Venky Shankar 
Signed-off-by: Kotresh HR 
Reviewed-on: http://review.gluster.org/11820
Reviewed-by: Milind Changire

geo-rep: Fix history failure

2015-08-06T05:49:52+00:00

Both ACTIVE and PASSIVE workers register to changelog
at almost same time. When PASSIVE worker becomes ACTIVE,
the start and end time would be current stime and register_time
repectively for history API. Hence register_time would be less
then stime for which history obviously fails. But it will
be successful for the next restart as new register_time > stime.

Fix is to pass current time as the end time to history call
instead of the register_time.

Also improvised the logging for ACTIVE/PASSIVE switching.

BUG: 1247882
Change-Id: I40c582cc32fe29a6c30340ec81a3b5d30e461e71
Reviewed-on: http://review.gluster.org/11524
Tested-by: Gluster Build System 
Tested-by: NetBSD Build System 
Reviewed-by: Aravinda VK 
Reviewed-by: Venky Shankar 
Signed-off-by: Kotresh HR 
Reviewed-on: http://review.gluster.org/11784
Reviewed-by: Milind Changire

geo-rep: Fix fd referenced before assignment

2015-07-13T02:48:02+00:00

Fix fd reference before assignment in mgmt_lock
function.

BUG: 1240607
Change-Id: Ib8a5207a516eecb55a5f906ff5d49abc98cab2a6
Reviewed-on: http://review.gluster.org/11318
Signed-off-by: Kotresh HR 
Reviewed-on: http://review.gluster.org/11563
Tested-by: Gluster Build System 
Tested-by: NetBSD Build System 
Reviewed-by: Saravanakumar Arumugam 
Reviewed-by: Milind Changire 
Reviewed-by: Venky Shankar

geo-rep: ignore ESTALE as ENOENT

2015-06-28T14:47:40+00:00

When DHT can't resolve a File it raises ESTALE, ignore ESTALE errors
same as ENOENT after retry.

Affected places:
    Xattr.lgetxattr
    os.listdir
    os.link
    Xattr.lsetxattr
    os.chmod
    os.chown
    os.utime
    os.readlink

BUG: 1236093
Change-Id: I53f8dfa47911da93e0dcc20213afcbb47a14ccd8
Reviewed-On: http://review.gluster.org/11296
Original-Author: Aravinda VK 
Signed-off-by: Kotresh HR 
Reviewed-on: http://review.gluster.org/11430
Tested-by: NetBSD Build System 
Reviewed-by: Milind Changire 
Reviewed-by: Vijay Bellur

geo-rep: Fix geo-rep fanout setup with meta volume

2015-06-25T17:27:28+00:00

Lock filename was formed with 'master volume id'
and 'subvol number'. Hence multiple slaves try
acquiring lock on same file and become PASSIVE
ending up not syncing data. Using 'slave volume id'
in lock filename will fix the issue making lock
file unique across different slaves.

Change-Id: I64c84670a5d9e1b0dfbdeb4479ee6b8e0c6b829e
BUG: 1234898
Reviewed-On: http://review.gluster.org/11367
Signed-off-by: Kotresh HR 
Reviewed-on: http://review.gluster.org/11366
Tested-by: Gluster Build System 
Tested-by: NetBSD Build System 
Reviewed-by: Saravanakumar Arumugam 
Reviewed-by: Venky Shankar

geo-rep: Fix toggling of use_meta_volume config

2015-06-25T17:15:41+00:00

If meta-volume is deleted and use_meta_volume
is set to false, geo-rep still fails complaining
meta volume is not mounted. The patch fixes that
issue.

BUG: 1234695
Change-Id: I5ec7ad85399b90e85bca1d36750dd3fb29222db9
Reviewe-On: http://review.gluster.org/11358
Signed-off-by: Kotresh HR 
Reviewed-on: http://review.gluster.org/11359
Tested-by: NetBSD Build System 
Tested-by: Gluster Build System 
Reviewed-by: Saravanakumar Arumugam 
Reviewed-by: Venky Shankar

geo-rep: Ignore ESTALE during unlink/rmdir

2015-06-01T06:38:15+00:00

during unlink/rmdir of Parent_GFID/Basename, if parent
directory does not exists. Parent GFID will not get resolved
and DHT raises ESTALE instead of ENOENT.

Now ESTALE errors ignored during unlink/rmdir

BUG: 1223286
Change-Id: If275c89fb9fc7d16004550805a4cd65be818540d
Signed-off-by: Aravinda VK 
Reviewed On: http://review.gluster.org/#/c/10837/
Reviewed-on: http://review.gluster.org/10913
Tested-by: NetBSD Build System 
Tested-by: Gluster Build System 
Reviewed-by: Kotresh HR 
Reviewed-by: Venky Shankar

geo-rep: Fix Data counter issue in status

2015-05-31T14:44:20+00:00

ENTRY and META operations executed sequentially, DATA operations
are handled async, increment happens when a changelog parsed.
Decrement happens after the sync of all files.

files_in_batch was reset multiple times in batch instead of once.

BUG: 1224100
Change-Id: I87617f2fd5f4d3221a1c9f9d5a8efb0686c42bbe
Signed-off-by: Aravinda VK 
Reviewed On: http://review.gluster.org/#/c/10911/
Reviewed-on: http://review.gluster.org/10912
Tested-by: Gluster Build System 
Tested-by: NetBSD Build System 
Reviewed-by: Venky Shankar