This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
How to kick process out of uninterruptible sleep?
- From: Riccardo Murri <riccardo dot murri at uzh dot ch>
- To: systemtap at sourceware dot org
- Date: Wed, 9 Nov 2016 14:36:06 +0000 (UTC)
- Subject: How to kick process out of uninterruptible sleep?
- Authentication-results: sourceware.org; auth=none
- Reply-to: riccardo dot murri at uzh dot ch
Hello,
Is the following possible at all with systemtap?
I have a server where a process got stuck in "uninterruptible sleep" during NFS
access. Why it got stuck is unclear (possibly a kernel oops
happening at a bad time), but the NFS server and client look fine and I can
normally access any file that /proc/PID/fd reports as open. The server looks
otherwise fine and other processes run OK.
`ps -O wchan` shows it's stuck in `nfs_wait_bit_uninterruptible`, but inpection
of the kernel sources does not give much help[1].
# ps -O wchan:32 33033
PID WCHAN S TTY TIME COMMAND
33033 nfs_wait_bit_uninterruptible D ? 00:43:32 ...
Is there a tapset that can abort the syscall this process is stuck in? We know
from the logfiles that it has terminated its job, so it's no problem if it ends
with an error.
[1]: http://lxr.linux.no/#linux+v3.10/fs/nfs/pagelist.c#L261
Thanks for any hint!
Riccardo
--
Riccardo Murri, Anna-Heer-Strasse 10, CH-8057 Zürich, Switzerland