[systemd-devel] systemd 215: NFS mount marked successfully even if it's not (and other NFS mounts issues)

[Sandro Tosi]

Hey!

we are seeing some weird behavior of systemd 215 (I know it's a rather
old version, but it's the one on Debian Jessie, so it has quite a big
audience) with NFS mounts. We had a lengthy and helpful discussion[1]
with the Debian systemd maintainers (CC'ed) about that and in one of
the last messages[2] we spotted something that sounds really weird and
we'd like you to have a look:

Feb 23 06:50:43 SERVER systemd[1]: Mounting /mnt/NFSSERVER1_VOL...
Feb 23 06:50:43 SERVER systemd[1]: About to execute: /bin/mount -n
XXX.YYY.32.75:/vol/vol3 /mnt/NFSSERVER1_VOL -t nfs -o
ro,intr,nolock,tcp,rdirplus,noatime,_netdev
Feb 23 06:50:43 SERVER systemd[1]: mnt-NFSSERVER1_VOL.mount changed
dead -> mounting
Feb 23 06:50:43 SERVER systemd[574]: Executing: /bin/mount -n
XXX.YYY.32.75:/vol/vol3 /mnt/NFSSERVER1_VOL -t nfs -o
ro,intr,nolock,tcp,rdirplus,noatime,_netdev
Feb 23 06:52:13 SERVER systemd[1]: mnt-NFSSERVER1_VOL.mount mounting
timed out. Stopping.
Feb 23 06:52:13 SERVER systemd[1]: mnt-NFSSERVER1_VOL.mount changed
mounting -> mounting-sigterm
Feb 23 06:52:13 SERVER systemd[1]: Child 574 belongs to mnt-NFSSERVER1_VOL.mount
Feb 23 06:52:13 SERVER systemd[1]: mnt-NFSSERVER1_VOL.mount mount
process exited, code=killed status=15
Feb 23 06:52:13 SERVER systemd[1]: mnt-NFSSERVER1_VOL.mount changed
mounting-sigterm -> mounted
Feb 23 06:52:13 SERVER systemd[1]: Job mnt-NFSSERVER1_VOL.mount/start
finished, result=done
Feb 23 06:52:13 SERVER systemd[1]: Mounted /mnt/NFSSERVER1_VOL.
Feb 23 06:52:13 SERVER systemd[1]: Starting Remote File Systems.
Feb 23 06:52:13 SERVER systemd[1]: remote-fs.target changed dead -> active
Feb 23 06:52:13 SERVER systemd[1]: Job remote-fs.target/start
finished, result=done
Feb 23 06:52:13 SERVER systemd[1]: Reached target Remote File Systems.
Feb 23 06:52:13 SERVER systemd[1]: mnt-NFSSERVER1_VOL.mount changed
mounted -> failed

so it appears the mount is marked as mounted, while there was actually an error.

We started the conversation at [1] with other NFS issues: some mounts
were not mounted or appeared mounted twice, and they might all be
related to some underlying problem with NFS shares and (maybe) this
specific version of systemd or some other systems component, and your
help in debugging and fixing them is greatly appreciated!

we are willing to run tests or change configuration as needed, but it
might be a bit difficult to upgrade to a newer version of systemd.

Attached is the first few minutes of boot of the machine from where we
extracted the log above (it's anonymized, but it should preserve all
the useful information), and let me know if you need more details
(some of them are in the thread at [1]) or some config changes.

[1] http://lists.alioth.debian.org/pipermail/pkg-systemd-maintainers/2016-January/thread.html#10382
+ http://lists.alioth.debian.org/pipermail/pkg-systemd-maintainers/2016-February/thread.html#10734
[2] http://lists.alioth.debian.org/pipermail/pkg-systemd-maintainers/2016-February/010870.html

Thanks a ton in advance!

Please keep me (and the Debian systemd maints) in CC, as I'm not
subscribed to this list

Cheers,
-- 
Sandro "morph" Tosi
My website: http://sandrotosi.me/
Me at Debian: http://wiki.debian.org/SandroTosi
G+: https://plus.google.com/u/0/+SandroTosi
-------------- next part --------------
A non-text attachment was scrubbed...
Name: boot.log.gz
Type: application/x-gzip
Size: 32205 bytes
Desc: not available
URL: <https://lists.freedesktop.org/archives/systemd-devel/attachments/20160310/bbf1285a/attachment-0001.bin>