glusterfs.git/xlators/performance, branch v3.3.0qa39

core: coverity issues fixed

2012-04-23T21:52:57+00:00

this is not a complete set of issues getting fixed. Will
address other issues in another patch.

Change-Id: Ib01c7b11b205078cc4d0b3f11610751e32d14b69
Signed-off-by: Amar Tumballi 
BUG: 789278
Reviewed-on: http://review.gluster.com/3145
Tested-by: Gluster Build System 
Reviewed-by: Jeff Darcy

clean up handling of special client pids

2012-04-23T21:50:55+00:00

- no need to name a lower boundary [drop GF_CLIENT_PID_MIN]
- drop utility function to decide about pid speciality [gf_pid_client_check()];
  when we implement uniform policy for special pids, use just comparison with
  GF_CLIENT_PID_MAX (currently such uniform policy exists in io-threads, where
  special clients get lowest priority)
- io-threads: drop numeric literal, use GF_CLIENT_PID_MAX
- xattr flipping (in fuse):
  - fuse_xattr_alloc_default() made static
  - avoid compiler warnings wrt. partial enum traversal in the switch
    by not casting the original numeric value to gf_client_pid_t (enum) type
  - trim down the switch to those cases with which there is something to do
    [GF_CLIENT_PID_GSYNCD, GF_CLIENT_PID_HADOOP]
  - eliminate the (theoretical) chance of succesfully returning
    from fuse_flip_xattr_ns() with unallocated nkey
    [as was the case with pids GF_CLIENT_PID_{MIN,MAX,DEFRAG}]

Change-Id: Ic2b83b908d8e3be23a7fba1e5895869828a6194b
BUG: 765433
Signed-off-by: Csaba Henk 
Reviewed-on: http://review.gluster.com/3170
Tested-by: Gluster Build System 
Reviewed-by: Jeff Darcy 
Reviewed-by: Anand Avati

performance/io-threads: Change the thread scaling logic.

2012-04-20T16:15:29+00:00

Change-Id: Idc0db1cb2c26dcb0b36a58d4d43f2da10cde62c5
BUG: 812515
Signed-off-by: Pranith Kumar K 
Reviewed-on: http://review.gluster.com/3186
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

performance/quick-read: disable reading from cache if unlink has

2012-04-18T07:00:16+00:00

happened after fd was opened.

If dentry being unlinked is the last one, there is a possibility that
later, a new entry with same name being created. After creation of new
file, we shouldn't be returning the contents of this new file, when
application reads using fd opened before unlink.

Change-Id: Ic98aa9cd76d24e8a37ce02a0aae6c66290043192
BUG: 811976
Signed-off-by: Raghavendra G 
Reviewed-on: http://review.gluster.com/3151
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

md-cache: fix dict memory leak in mdc_inode_xatt_update

2012-04-12T21:21:39+00:00

We currently either dict_ref() or dict_copy() into mdc->xattr.
The double dict_ref() call path is obvious. Alternatively,
dict_copy() does not ref the dict in the situation where it
creates a new dict, but we only call dict_copy() in this
context when we already have a dict, which should mean the
extra ref is unnecessary.

BUG: 801582
Change-Id: I72e3f1a16f942f04752c033812d13d50ee7c2c2c
Signed-off-by: Brian Foster 
Reviewed-on: http://review.gluster.com/3138
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

md-cache: fix time comparison to support 0 timeout

2012-04-02T18:25:12+00:00

The time comparison currently causes caching when the current
time matches the timeout. This causes caching to occur even
when the timeout parameter is set to 0. Fix the condition to
support the no caching scenario. A longer term approach is to
use gettimeofday() for more accurate granularity, but this
apparently fixes a known bug.

Change-Id: I6552c5961743edb22a1aec2e6a3a6e041d0647b2
BUG: 809193
Signed-off-by: Brian Foster 
Reviewed-on: http://review.gluster.com/3066
Tested-by: Gluster Build System 
Reviewed-by: Jeff Darcy

performance/quick-read: open shouldn't be worried about checking

2012-04-02T09:38:24+00:00

whether it is being done on a deleted dentry.

lookup would've taken care of sending ENOENT, in case the entry
was already deleted.

Change-Id: I79c51e34ae3a202eb04123a2a58d0e819b6e94fa
BUG: 808017
Signed-off-by: Raghavendra G 
Reviewed-on: http://review.gluster.com/3044
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

performance/io-cache: Enhance state-dump output

2012-04-01T13:16:31+00:00

Along with weight, print uuid and path. If only the inode is
cached, only weight, path and uuid will be printed.

Change-Id: Ie7567832915319f26c6bab8a34576f6c43be7998
BUG: 806900
Signed-off-by: shishir gowda 
Reviewed-on: http://review.gluster.com/3014
Tested-by: Gluster Build System 
Reviewed-by: Amar Tumballi 
Reviewed-by: Vijay Bellur

performance/quick-read: fix build errors.

2012-03-29T06:02:52+00:00

Change-Id: I8c4781bdd6c3cfd50cff95c31ecd6d653b916ac3
BUG: 765363
Signed-off-by: Raghavendra G 
Reviewed-on: http://review.gluster.com/3028
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

performance/quick-read: make unlink aware of open-behind behaviour of quick-read.

2012-03-28T17:40:41+00:00

quick-read should do an open of all the fds that are waiting for to be
opened behind before doing an unlink on a path.

Change-Id: I4c8d65c0511e8a37765ff57e30d99dd89b56c4fc
BUG: 765363
Signed-off-by: Raghavendra G 
Reviewed-on: http://review.gluster.com/506
Tested-by: Gluster Build System 
Reviewed-by: Shishir Gowda 
Reviewed-by: Vijay Bellur