lkml.org 
[lkml]   [1997]   [Oct]   [1]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Patch for 2.1.57 lockd -- please test!
Gary White (Network Administrator) wrote:
> Get this messags while unmounting.
>
> moe:/# mount ringo:/ /mnt
> moe:/# umount ringo:/
> lockd_down: waiting for pid 234 to exit
> lockd_down clearing pid
>
> But the pid does not clear.

The messages indicate that the lockd process was somehow hung and failed
to respond to the kill signal. The "clearing pid" message just means
that the slot has been cleared so that a new lockd can be created by the
next mount. If the lockd process refuses to exit, lockd_down is pretty
much forced to clear the slot; otherwise no future mount would get a
lockd.

Having the lockd processes getting hung is not good. Which NFS server
are you using?

> 179 1 DW 0:00 (lockd)
> 182 1 DW 0:00 (lockd)
> 185 1 DW 0:00 (lockd)
> 188 1 DW 0:00 (lockd)
> 191 1 DW 0:00 (lockd)
> 194 1 DW 0:00 (lockd)
> 197 1 DW 0:00 (lockd)
> 202 1 DW 0:00 (lockd)
> 234 1 DW 0:00 (lockd)
>
> Did several mounts and umounts, and waited 15-20 minutes and pids never
> clears. Load average still contnues to climb although does not effect
> response.

Could you check whether multiple lockd's are created when you just
mount, before you've done any unmounts? There seem to be two (or more)
problems in lockd --
(1) it was allowing multiple lockd's to be created at mount time (which
my patch is intended to fix), and
(2) lockd processes are getting hung in the DW state, which I haven't
tracked down yet, but which may be server-related.

Thanks for helping check this out ..

Regards,
Bill

\
 
 \ /
  Last update: 2005-03-22 13:40    [W:0.075 / U:0.192 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site