ZFS lockup in "zfs" state

[Ivan Voras]

Jeremy Chadwick wrote:
> On Wed, Feb 13, 2008 at 11:43:42AM +0100, Ivan Voras wrote:
>> Hi,
>>
>> A machine just wedged its ZFS file systems (UFS file systems were still
>> running), all unresponsive processes were in "zfs" state. This is
>> 7.0-RC1 on i386 with 2 GB RAM. It has happened at least once before.
>>
>> I have a kernel core dump (dumped from kdb/ddb on hotkey), if anyone's
>> interested.
> 
> Were you copying any data from a ZFS pool to a UFS filesystem, or
> ZFS-to-ZFS, at that time? 

I don't know - the machine is a server that's been doing it's job (web
server, mysql, java). Depending on when the problem started, there could
have been a rsync from ZFS to UFS for backup purposes.

> Does your coredump's backtrace look at all
> similar to the below report?
> 
> http://lists.freebsd.org/pipermail/freebsd-stable/2008-January/040047.html

I vaguely remember there being something about sockbuf when I had the
online debugger running and java (tomcat) is definitely running on the
server. Can you advise me how to extract this kind of information from a
kernel core dump (kgdb doesn't have show alllocks, etc :) )?

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: OpenPGP digital signature
Url : http://lists.freebsd.org/pipermail/freebsd-fs/attachments/20080213/e2441a80/signature.pgp