glusterfs.git, branch dht-stale-layout-fixes

change branch name.

2014-08-09T04:45:08+00:00

Change-Id: I4790d5a39f5e3d3337c50ba74b87ca1e9ab99e75
Signed-off-by: Raghavendra G 
Reviewed-on: http://review.gluster.org/8445

Wrong test in self-heal.t

2014-08-08T05:01:56+00:00

self-heal.t checks for gfid before and after an operation, but fails
to do it on the same file since the prefix is not the same: undefined
$B1 before and $B0 after.

The errorcause the test to fail on NetBSD. I do not know how this test
could pass on Linux before.

BUG: 764655
Change-Id: I6df1175e177855afca1736cf8c847c51fe407576
Signed-off-by: Emmanuel Dreyfus 
Reviewed-on: http://review.gluster.org/8431
Reviewed-by: Ravishankar N 
Reviewed-by: Pranith Kumar Karampuri 
Tested-by: Pranith Kumar Karampuri

build: make GLUSTERD_WORKDIR rely on localstatedir

2014-08-07T08:17:29+00:00

- Break-way from '/var/lib/glusterd' hard-coded previously,
  instead rely on 'configure' value from 'localstatedir'
- Provide 's/lib/db' as default working directory for gluster
  management daemon for BSD and Darwin based installations
- loff_t is really off_t on Darwin
- fix-off the warnings generated by clang on FreeBSD/Darwin
- Now 'tests/*' use GLUSTERD_WORKDIR a common variable for all
  platforms.
- Define proper environment for running tests, define correct PATH
  and LD_LIBRARY_PATH when running tests, so that the desired version
  of glusterfs is used, regardless where it is installed.
  (Thanks to manu@netbsd.org for this additional work)

Change-Id: I2339a0d9275de5939ccad3e52b535598064a35e7
BUG: 1111774
Signed-off-by: Harshavardhana 
Signed-off-by: Emmanuel Dreyfus 
Reviewed-on: http://review.gluster.org/8246
Tested-by: Gluster Build System

Regression test portability: truncate, md5

2014-08-06T11:49:14+00:00

Add shell functions to replace truncate and md5, which are Linux specific
Resubmit because of failed regression test with no apparent cause

BUG: 764655
Change-Id: I07200cf886bd52904a5cf63c66f43f0b1cc91540
Signed-off-by: Emmanuel Dreyfus 
Reviewed-on: http://review.gluster.org/8341
Tested-by: Gluster Build System 
Reviewed-by: Harshavardhana 
Tested-by: Harshavardhana 
Reviewed-by: Vijay Bellur

tests: weighted-rebalance.t shouldn't write to '/dev/tty'

2014-08-06T11:39:20+00:00

On our jenkins instance "/dev/tty" doesn't exist, necessary
output fails as below message

~~~
./tests/features/weighted-rebalance.t: \
         line 72: /dev/tty: No such device or address
~~~

Comment out the debugging code

Change-Id: Iba29b80c8ba2dcaab3d6654d7c54332a915bffb8
BUG: 1114680
Signed-off-by: Harshavardhana 
Reviewed-on: http://review.gluster.org/8421
Tested-by: Gluster Build System 
Reviewed-by: Jeff Darcy 
Reviewed-by: Vijay Bellur

libglusterfs: wrong function definition of synclock_destory().

2014-08-06T08:29:28+00:00

synclock_destory() has been prototyped in syncop.h,
how-ever synclock_destroy() is the actual function used in syncop.c.
Correcting this function definition along with few typos.


Change-Id: I35a818190c1d37c303279ca7a820f01895751bd9
BUG: 1075417
Signed-off-by: Humble Chirammal 
Reviewed-on: http://review.gluster.org/8266
Tested-by: Gluster Build System 
Reviewed-by: Raghavendra Talur 
Reviewed-by: soumya k 
Reviewed-by: Poornima G 
Reviewed-by: Vijay Bellur

doc: afr-locks and its evolution

2014-08-05T10:44:53+00:00

History and current state of how AFR uses the inode and entry locks
provided by the locks xlator

Thanks to Pranith for patiently explaining this.

Change-Id: I25d21c6361e95aa0b3edcffe9b3b9e470f488bc6
BUG: 1118309
Signed-off-by: Ravishankar N 
Reviewed-on: http://review.gluster.org/8364
Tested-by: Gluster Build System 
Reviewed-by: Krutika Dhananjay 
Reviewed-by: Humble Devassy Chirammal 
Reviewed-by: Vijay Bellur

gluster vol set nfs-ganesha.enable off fails to unexport volume.

2014-08-05T10:33:40+00:00

 Due to a parameter type change in upstream nfs-ganesha
( v2.1 and above ), gluster vol set  nfs-ganesha.enable
 off doesn't unexport the volume in question if it was already
exported.
A minor change in the hook script would result in the
expected behaviour.

Change-Id: I46f76f3cde6b6ebd7a94d5ee16f05a3648a2840f
BUG: 1124711
Signed-off-by: Meghana Madhusudhan 
Reviewed-on: http://review.gluster.org/8390
Reviewed-by: soumya k 
Tested-by: Gluster Build System 
Reviewed-by: Humble Devassy Chirammal 
Reviewed-by: Vijay Bellur

socket: add boundary checks for iobuf_get2 over rpc_hdr_bytes

2014-08-05T01:54:53+00:00

A malformed packet can cause a OOM while peforming iobuf_get2()
on a large enough packet side.

Such a scenario is observed when running vulnerability tests,
it would look like one of those tests perhaps based on DDOS (Denial
of Service) attacks hand-crafts a RPC packet which is of a large
enough size - since we do not verify the size and do not provide
any boundary checks, there are secenarios where it leads to OOM.

Reproducible consistently with those tests has revealed that we
should be ideally be adding a boundary check.

Limit such an allocation to a 1gigabyte, as it might be sufficient
to handle for all variety of RPC packets.

Change-Id: I5f1411dd96d6f167993d28a1718ffef2fb4e9923
Signed-off-by: Harshavardhana 
Reviewed-on: http://review.gluster.org/8384
Reviewed-by: Raghavendra G 
Tested-by: Raghavendra G

Regression test portability: arequal-checksum

2014-08-05T00:41:19+00:00

Building arequal-checksum on non Linux systems requires a few adjustments:
- use __builtin_alloca() on all platforms
- on systems without argp, get it from contrib/standalone-argp
  (this requires adding cflags support to the build_tester function)
- FTW_ACTIONRETVAL option to nftw(3) is GNU libc specific, work around
  if it does not exist.
- md5sum is Linux-specific. Use md5 -n on NetBSD and md5 -q on FreeBSD/Darwin
- Use 'cc' as synonymous for all compilers, it can behave as gcc/clang depending
  on which is default
- cleanup tabs/whitespaces

BUG: 764655
Change-Id: I9090c17da596fbf00fc1fbd7593163ce8cd3b84c
Signed-off-by: Emmanuel Dreyfus 
Signed-off-by: Harshavardhana 
Reviewed-on: http://review.gluster.org/8283
Tested-by: Gluster Build System