[Gluster-users] crash in __socket_ioq_new

[Chetan Ahuja]

I posted this on irc (#gluster) a couple of times in the last few days but
got no response. Trying my luck here:

 <chetan> I'm seeing this crash in 2.0.1 server codebase
 #4  0x00007f6a42848a56 in free () from /lib/libc.so.6
 #5  0x00007f6a4157ff99 in __socket_ioq_new (this=<value optimized out>,
buf=0x1 <Address 0x1 out of ≠, len=<value optimized out>,
vector=0x7f6a3c019db0,   count=1006736528, iobref=0x0) at socket.c:313
 #6  0x00007f6a41581c08 in socket_event_handler (fd=5883, idx=5883,
data=0x6, poll_in=-1, poll_out=1116630944, poll_err=0) at socket.c:796

  This happened on  volume with 4 bricks. Two subvolumes of two replicas
each and the final volume distributed over the replicas. There are a large
set of clients writing to these volumes. The write patterns are such that
all clients are writing into different files. In other words, no two clients
are writing simultaneously to the same file at any time.

  I haven't seen this stack trace on either the mailing list archives or the
bugzilla and I don't see any code changes in the relevant code in 2.0.6
either.  If the developers would confirm that if it's a known bug and/or has
a known workaround... I'd really appreciate that.

Thanks a lot
Chetan