[Gluster-devel] Something wrong with release-3.6 branch

[Justin Clift]

Hi Vijay,

There's "something" wrong with the release-3.6 branch. ;)

Several of the Rackspace regression VM's have been somehow
killed over the last few hours, and needing to be rebuilt.

Not sure exactly what's being done to them, as they stop
responding to ssh, and the best I can do is get them into
rescue mode... where nothing is obviously showing up as
being wrong. (but they're still useless when booted
normally)

Anyway, I'm suspecting there's something in the release-3.6
branch that's causing it.

Here's a regression run of release-3.6 branch, with no
other CR's applied:

  http://build.gluster.org/job/rackspace-regression-2GB/638/console

Note in the compilation phase these two warnings:

  /home/jenkins/root/workspace/rackspace-regression-2GB/xlators/cluster/dht/src/dht-common.c: In function ‘dht_lookup_everywhere_done’:
  /home/jenkins/root/workspace/rackspace-regression-2GB/xlators/cluster/dht/src/dht-common.c:1229: warning: implicit declaration of function ‘dht_fill_dict_to_avoid_unlink_of_migrating_file’

Guessing these are significant. :)

Also notice that the regression testing phase proper is
broken pretty much from the start.  So, I think there's
something fundamentally busted in release-3.6 atm.

Here's the master branch, running on the same VM just a
bit before:

  http://build.gluster.org/job/rackspace-regression-2GB/636/console

It compiles fine, and the regression testing seems ok.
(I aborted it early due only to impatience :>)

Anyway, this is 100% reproducible, I rebooted that VM
between runs, cleaned out the workspace and other
artifacts.  So it shouldn't be due to anything silly
like that.

Any ideas?  Hopefully it's something simple to fix. :)

+ Justin

--
GlusterFS - http://www.gluster.org

An open source, distributed file system scaling to several
petabytes, and handling thousands of clients.

My personal twitter: twitter.com/realjustinclift