zfs discuss - Nov 2009 - ls -l hang, process unkillable

hello, 

one of my colleague has a problem with an application. the sysadmins,
responsible for that server told him that it was the applications fault, but i
think they are wrong, and so does he.

from time to time, the app gets unkillable and when trying to list the contents
of some dir which is being read/written by the app, "ls" can list the
contents, but "ls -l" gets stuck and simply hangs. cp,rm,mv on some
file in that dir doesn`t work either.

i think this is a solaris/kernel/zfs problem.

can somebody give a hint how to analyze/fix this? 

i can provide more input on request.

regards
roland
-- 
This message posted from opensolaris.org

roland wrote:
> hello, 
> 
> one of my colleague has a problem with an application. the sysadmins,
responsible for that server told him that it was the applications fault, but i
think they are wrong, and so does he.
> 
> from time to time, the app gets unkillable and when trying to list the
contents of some dir which is being read/written by the app, "ls" can
list the contents, but "ls -l" gets stuck and simply hangs. cp,rm,mv
on some file in that dir doesn`t work either.
> 
> i think this is a solaris/kernel/zfs problem.
> 
> can somebody give a hint how to analyze/fix this? 
Start with something like:

pgrep ls

pstack <PID of ls>

echo "0t<PID of ls>::pid2proc|::walk thread|::findstack -v" |
mdb -k

victor

thanks.

we will try that if the error happens again - needed to reboot as a quick-fix,
as the machine is in production....

regards
roland
-- 
This message posted from opensolaris.org