glusterfs.git/xlators/cluster/dht/src/dht-common.c, branch v3.1.0qa43

distribute: check for 'conf' before dereferencing it

2010-10-05T09:48:14+00:00

Signed-off-by: Amar Tumballi 
Signed-off-by: Vijay Bellur 

BUG: 1806 ()
URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=1806

Change GNU GPL to GNU AGPL

2010-10-04T14:50:05+00:00

Signed-off-by: Pranith Kumar K 
Signed-off-by: Vijay Bellur 

BUG: 1388 ()
URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=1388

rmdir: introduce extra flags parameter in FOP prototype

2010-10-03T05:18:56+00:00

Signed-off-by: Anand V. Avati 
Signed-off-by: Vijay Bellur 

BUG: 971 (dynamic volume management)
URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=971

stat enhancements

2010-10-01T16:33:50+00:00

* dht to send 'setxattr' to all subvolumes in the layout
* server dumps info on total bytes read/written for 'trusted.io.stat.dump' key
* server dumps all the mount point IP for 'trusted.list.mount.point' key.
* io-stats dumps latency information only if measured

Signed-off-by: Amar Tumballi 
Signed-off-by: Vijay Bellur 

BUG: 1701 (better statistics  gathering in glusterd)
URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=1701

Changes to replace flock with gf_flock across GlusterFS.

2010-10-01T12:05:53+00:00

Signed-off-by: Pavan Vilas Sondur 
Signed-off-by: Vijay Bellur 

BUG: 865 (Add locks recovery support in GlusterFS)
URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=865

distribute: Propagate -1 op_ret on failed fop

2010-09-28T11:50:48+00:00

Signed-off-by: Shehjar Tikoo 
Signed-off-by: Vijay Bellur 

BUG: 1717 (dht_attr_cbk does not propagate op_ret on failed fop causing nfs crash)
URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=1717

distribute: don't check for inode's gfid in _cbk

2010-09-28T05:53:21+00:00

* in dht_lookup_dir_cbk(), at the entry itself we are dereferencing
  local->inode, which can be NULL in many cases. Hence no need to
  check/dereference that variable. Just check 'local->gfid'.

Signed-off-by: Amar Tumballi 
Signed-off-by: Vijay Bellur 

BUG: 1696 ()
URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=1696

distribute: don't update the inode's gfid directly

2010-09-23T14:51:06+00:00

Signed-off-by: Amar Tumballi 
Signed-off-by: Vijay Bellur 

BUG: 1680 ()
URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=1680

distribute: bring in feature to fix the layout of root inode

2010-09-23T14:51:02+00:00

* If the errno is ESTALE for root, fuse doesn't send fresh lookup on
  the path (it does for all other inodes). Earlier layout mismatch for
  root never used to happen inside codebase, and if noticed, it was
  serious issue.

  Now with DVM, after a add-brick or a remove-brick, layout can be
  changed even for 'root' inode. Needed to fix this issue inside
  distribute itself, instead of fuse sending 'fresh' lookup on root.

Signed-off-by: Amar Tumballi 
Signed-off-by: Vijay Bellur 

BUG: 1630 ()
URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=1630

distribute: while selfhealing directory, send proper gfid in dict

2010-09-22T15:09:33+00:00

* this was the root cause for having layout mismatches in case of
  add-brick, because the gfid of directories on newly added brick
  was always mismatching, which caused many operation on that
  particular brick fail.

Signed-off-by: Amar Tumballi 
Signed-off-by: Vijay Bellur 

BUG: 1629 (files missing during add-brick)
URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=1629