glusterfs.git, branch v3.2.5qa8

cluster/distribute lookup: Request layout info for dir's

2011-11-10T11:08:34+00:00

Change-Id: I17e40beb2c6e77a802fb987bed6f0e4be8010ef6
BUG: 3803
Reviewed-on: http://review.gluster.com/707
Tested-by: Gluster Build System 
Reviewed-by: Amar Tumballi

cluster/afr: Handle split-brain/all-fool xattrs for directory

2011-11-09T17:57:20+00:00

Change-Id: I058ed91494e1a9bbcf5e6e6c49e8ee4f7e014e23
BUG: 3796
Reviewed-on: http://review.gluster.com/695
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

cluster/distribute stat: Take ref on inode before to prevent crash.

2011-11-09T17:32:34+00:00

Possible situation is a forget call which unref's the inode, and the
layout gets freed before we access it in stat.
By taking a ref before we call ctx get, the layout is going to be
persistent till we unref it.

Also use local call_cnt var instead of layout->cnt when wind is
involved

Change-Id: Ie04a7afd9e2c89cc14c08dd923d06c5f0193beb8
BUG: 3730
Reviewed-on: http://review.gluster.com/687
Tested-by: Gluster Build System 
Reviewed-by: Amar Tumballi

cluster/distribute lookup: send revalidate calls to all subvols for directories

2011-11-09T17:27:58+00:00

If mkdir fails on a subvolume, layout is set taking into account
only the subvols where it was successful. stat does not trigger
selfheal, as its layout based.
Revalidate on directories needs to be sent to all subvols, to fix
the error, and not just on the layout.

Change-Id: Ia3ce2e5a042b2d55cddd7bd8cf31a94de6023234
BUG: 3793
Reviewed-on: http://review.gluster.com/688
Tested-by: Gluster Build System 
Reviewed-by: Amar Tumballi

stripe read: Wipe stat in reply to prevent truncated error.

2011-11-09T06:11:48+00:00

If the read request, does not fall to the subvolume with the largest
file size set, then we never return the correct size. This leads to
clients seeing a truncated file error.

The work around is to wipe stat being returned as part of read call.

Change-Id: I4e76d09d7919fd0be616fe140e08f89a0dbcc444
BUG: 3774
Reviewed-on: http://review.gluster.com/682
Tested-by: Gluster Build System 
Reviewed-by: Amar Tumballi

cluster/replicate: no more 'itransform' required for inode numbers

2011-11-09T06:10:05+00:00

as posix layer itself sends inode numbers based on gfid now

Change-Id: I184d4598cf541eadb0471c39e598f14d6ac9390a
BUG: 3042
Reviewed-on: http://review.gluster.com/632
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

cluster/afr: Handle absence of gfid in lookup

2011-11-05T12:56:28+00:00

Change-Id: I57fe8db82425a064ca01b23a47baca297c67560c
BUG: 3783
Reviewed-on: http://review.gluster.com/676
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

cluster/afr: unwind instead of returning directly if no open is required on any child

2011-11-04T16:55:31+00:00

Currently if no open is required on any child of replicate, we are
returning directly instead of unwinding and returning. This is leading
to the frame loss, which results in application hang. Unwind and then
return if no open is to be done on any child.

Change-Id: Ib59535b63751d9e98a4ceca39b4b73b14d850680
BUG: 3443
Reviewed-on: http://review.gluster.com/674
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

dht,afr: Fixed gfid problems

2011-11-04T16:54:01+00:00

*) removed uuid_generate usage in pump and afr
*) filled the gfids for the fops which were sending no gfid in loc

Change-Id: Id6780da76c3ae18ae5130368b687ea026e338ac0
BUG: 3760
Reviewed-on: http://review.gluster.com/673
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

libglusterfs: copy the gfid in entry_copy

2011-11-04T15:55:54+00:00

Change-Id: I8a43b5fbe7a90344f490090df853d47b651bc0ff
BUG: 3760
Reviewed-on: http://review.gluster.com/672
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur