glusterfs.git/xlators/cluster/dht/src, branch v3.4.0qa7

cluster/distribute: get_layout should account only available subvols

2013-01-24T07:43:39+00:00

The earlier logic used to check if (layout-spread-count <= subvol_cnt -
decommissioned bricks). With this if a subvol was down, and layout-spread was >
upsubvols, a mkdir ended up creating holes in the layout.

The fix is to consider only the combination of subvols which are usable (not
down or not decommissioned).

Change-Id: I61ad3bcaf4589f5a75f7887cfa595c98311ae3bb
BUG: 902610
Signed-off-by: shishir gowda 
Reviewed-on: http://review.gluster.org/4412
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

glusterd/cli: Updated the options descriptions for "volume set help"

2013-01-22T07:03:57+00:00

Change-Id: I0db00b7334bb9707ab48bd661ac03a3ad818d6e4
BUG: 893458
Signed-off-by: Avra Sengupta 
Reviewed-on: http://review.gluster.org/4393
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

core: fixes for gcc's '-pedantic' flag build

2013-01-22T06:17:36+00:00

* warnings on 'void *' arguments
* warnings on empty initializations
* warnings on empty array (array[0])

Change-Id: Iae440f54cbd59580eb69f3ecaed5a9926c0edf95
BUG: 875913
Signed-off-by: Avra Sengupta 
Reviewed-on: http://review.gluster.org/4219
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

cluster/distribute: If cached_subvol is down, return ENOTCONN in lookup

2013-01-21T20:03:03+00:00

When we follow a linkfile, and the lookup returns a ENOTCONN error, return
the error, as the cached subvol is down, and lookup_everywhere wont succeed,
but actually ends up clearing the linkfile, and clearing the namespace.

Change-Id: I772bf71531bc646e8fb62d3e8549a5fe0f3896da
BUG: 893378
Signed-off-by: shishir gowda 
Reviewed-on: http://review.gluster.org/4383
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

cluster/dht: update ctx-time only if we receive the new iatt

2013-01-17T08:26:57+00:00

1. Used local->postparent(contains merged iatt of all succesful calls) instead
of postparent for dht ctx time update.

2. dht_inode_ctx_time_update avoided in case of opret -1.

Change-Id: Ie04a7842a41c241f911b6a3f76267b996d27fb43
BUG: 881013
Signed-off-by: Varun Shastry 
Reviewed-on: http://review.gluster.org/4338
Reviewed-by: Shishir Gowda 
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

cluster/dht: Add "afr.readdir-failover=off" option the rebalance process

2012-12-17T03:45:53+00:00

By failing over readdir (default behaviour), rebalance could get duplicate
files, as readdir would re-read from offset 0. Rebalance should not attempt
to migrate these files again.

Additionally, we need to handle these cases as failure in rebalance crawl.

No test case provided, as we cannot determine the read child in afr.

Change-Id: If07508b4f92dacc17e0f695b48a866c7c66004be
BUG: 859387
Signed-off-by: shishir gowda 
Reviewed-on: http://review.gluster.org/4300
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

cluster/distribute: re-set layouts to prevent overlaps

2012-12-12T00:05:50+00:00

When subvols-per-directory option is used, with bricks addition/removal
the layouts might get distributed to other subvols, which were not part
of the layout before. We need to clean up layouts on old subvolumes, to
prevent overlaps.

Also, we need to make sure if layout-cnt is never less
than subvolume-cnt.

Change-Id: I00994a092ca0c99aedcc41bd9412d43460f88a04
BUG: 884455
Signed-off-by: shishir gowda 
Reviewed-on: http://review.gluster.org/4281
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

dht: support auto-NUFA option

2012-12-04T22:42:14+00:00

Many people have asked for behavior like the old NUFA, which builds and
seems to run but was previously impossible to enable/configure in a
standard way.  This change allows NUFA to be enabled instead of DHT from
the command line, with automatic selection of the local subvolume on each
host.

Change-Id: I0065938db3922361fd450a6c1919a4cbbf6f202e
BUG: 882278
Signed-off-by: Jeff Darcy 
Reviewed-on: http://review.gluster.org/4234
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

fix memory leaks

2012-12-04T20:14:28+00:00

* write-behind: free the inode context in wb_forget
* distribute: in readdirp callback put the allocated context to the inode
* distribute: check if the layout is NULL before accessing it in layout_unref

Change-Id: I7698f81b85b99d06bf6b01fc1a6e51e1593b5e27
BUG: 790709
Signed-off-by: Raghavendra Bhat 
Reviewed-on: http://review.gluster.org/4250
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

cluster/dht: fail fix-layout if any of the subvol is down

2012-11-30T06:11:06+00:00

If any subvolume is down, and a layout is re-written and hash
values change, entry names in the downed subvol can be reused
in the other subvol which got the same hash range. when the
downed subvol is brought back up, duplicate entried might appear

Also separated handling of ENOSPC and ENOTCONN error.

Change-Id: I5ed93990425a4cee70df2dab7c7c119fdc87ad56
BUG: 860663
Signed-off-by: shishir gowda 
Reviewed-on: http://review.gluster.org/4000
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati