Fwd: multiple processes hanging doing xfs ops -- 3.17.2

[L Walsh]

I just upped from 3.16.2 to 3.17.2 this morning, and after about 2 hours
uptime, got
several procs hanging in what looks like xfs territory.

Watched load slowing climb to about 16 as I was trying to bring it down
'gracefully'..

Last message looked like a disk controller message -- maybe it blocked other
processes, except sendmail should be running off a different controller
(built-in
controller handling system disks), while other procs would have been
accessing
2nd controller -- actually it looks like xfsdump was trying to run too,
so it could have
been accessing both controllers (backup disks are on 2nd controller, but
if backing
up system disks...)..

Attached are the call-traces from the hung tasks before the system came
down,


Also attached is my .config file.

Looks a bit peculiar ... any ideas?









-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: lines
URL: <http://oss.sgi.com/pipermail/xfs/attachments/20141112/432fe0d1/attachment-0001.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: config.txt
URL: <http://oss.sgi.com/pipermail/xfs/attachments/20141112/432fe0d1/attachment-0001.txt>