FS#48248 - [systemd] libreoffice-still hangs on start with systemd 229-2

Attached to Project: Arch Linux
Opened by Guillermo Leira (gleira) - Friday, 19 February 2016, 09:05 GMT
Last edited by Dave Reisner (falconindy) - Saturday, 18 June 2016, 13:03 GMT
Task Type Bug Report
Category Packages: Core
Status Closed
Assigned To Dave Reisner (falconindy)
Architecture x86_64
Severity Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Description:

After upgrading to systemd 229-2, libreoffice-still 5.0.5-1 no longer starts. It hangs on the splash screen.
Downgrading systemd to 228-4 solves the problem.
I have also downgraded systemd-sysvcompat and libsystemd.

Steps to reproduce:
Upgrade systemd, start libreoffice-still-5.0.5-1, downgrade systemd.
This task depends upon

Closed by  Dave Reisner (falconindy)
Saturday, 18 June 2016, 13:03 GMT
Reason for closing:  No response
Comment by Doug Newgard (Scimmia) - Friday, 19 February 2016, 15:26 GMT
Logs?
Comment by Guillermo Leira (gleira) - Saturday, 20 February 2016, 21:43 GMT
What logs can I provide? What do you need?
Comment by Andreas Radke (AndyRTR) - Sunday, 21 February 2016, 08:29 GMT
System journal and console output when starting from there.
Comment by Guillermo Leira (gleira) - Sunday, 21 February 2016, 13:29 GMT
[gleira@guillelinux ~]$ libreoffice
Warning: failed to read path from javaldx
^C
[gleira@guillelinux ~]$

It stays there until I press CTRL-C.

And the same error in journalctl:

feb 21 14:09:46 guillelinux libreoffice-startcenter.desktop[2757]: Warning: failed to read path from javaldx

No other relevant lines.


Comment by Guillermo Leira (gleira) - Sunday, 21 February 2016, 19:20 GMT
I have tried to enable java in Libreoffice, Options, Advanced. Now, when I start libreoffice from the command line, I get no errors or warnings. Nothing. But it also freezes in splash screen.

If I rename the .config/libreoffice/4 folder, I can start libreoffice.

If I clear the recent files list, libreoffice starts fine.

It seems that the problem is not related to Libreoffice. I downgraded systemd as soon as I noticed the problem at libreoffice, so I didn't notice that I can't browse using nautilus or thunar. Both of them hang on start for a few minutes. Nautilus crashes, Thunar freezes.

I have some cifs mounts defined in /etc/fstab with x-systemd.automount. If I comment all the ones that are not available at home (only at the office), everything goes fine. So it seems that the problem is some change in the way that this new version of systemd manages this feature.

This looks the same as always, it has to fail on non-available file servers.

dev-mqueue.mount loaded active mounted POSIX Message Queue File System
● mnt-Archivos_Madrid.mount loaded failed failed /mnt/Archivos_Madrid
● mnt-asusc.mount loaded failed failed /mnt/asusc
proc-fs-nfsd.mount loaded active mounted NFSD configuration filesystem

I tried some configurations, and it seems that the ones that are from other computers in the same network that are powered off are not a problem. The problem are the ones from the office.

These are the six lines on /etc/fstab (I have replaced the names of the servers and domains)

//PC.home-office.com/c$ /mnt/asusc cifs noauto,x-systemd.automount,credentials=/root/creds.casa,uid=gleira,gid=users,iocharset=utf8,nobrl,sec=ntlm 0 0
//PC.home-office.com/d$ /mnt/asusd cifs noauto,x-systemd.automount,credentials=/root/creds.casa,uid=gleira,gid=users,iocharset=utf8,nobrl,sec=ntlm 0 0
//PC.home-office.com/f$ /mnt/asusf cifs noauto,x-systemd.automount,credentials=/root/creds.casa,uid=gleira,gid=users,iocharset=utf8,nobrl,sec=ntlm 0 0

//nas.network-office.es/share1 /mnt/nas cifs noauto,x-systemd.automount,vers=1.0,credentials=/root/creds.pow,uid=gleira,gid=users,iocharset=utf8,nobrl,sec=ntlm 0 0
//fs01.network-office.es/share2 /mnt/gleira cifs noauto,x-systemd.automount,vers=2.1,credentials=/root/creds.pow,uid=gleira,gid=users,iocharset=utf8,nobrl 0 0
//fs01.network-office.es/share3 /mnt/Archivos_Madrid cifs noauto,x-systemd.automount,vers=2.1,credentials=/root/creds.pow,uid=gleira,gid=users,iocharset=utf8,nobrl 0 0

If I try to access /mnt/asusc, I get this in journalctl

feb 21 20:03:52 guillelinux systemd[1]: mnt-asusc.automount: Got automount request for /mnt/asusc, triggered by 3970 (ls)
feb 21 20:03:52 guillelinux systemd[1]: Mounting /mnt/asusc...
feb 21 20:03:58 guillelinux mount[3971]: Unable to find suitable address.
feb 21 20:03:58 guillelinux systemd[1]: mnt-asusc.mount: Mount process exited, code=exited status=32
feb 21 20:03:58 guillelinux systemd[1]: Failed to mount /mnt/asusc.
feb 21 20:03:58 guillelinux systemd[1]: mnt-asusc.mount: Unit entered failed state.
feb 21 20:04:00 guillelinux systemd[1]: mnt-asusc.automount: Got automount request for /mnt/asusc, triggered by 3979 (ls)
feb 21 20:04:00 guillelinux systemd[1]: Mounting /mnt/asusc...
feb 21 20:04:04 guillelinux mount[3980]: Unable to find suitable address.
feb 21 20:04:04 guillelinux systemd[1]: mnt-asusc.mount: Mount process exited, code=exited status=32
feb 21 20:04:04 guillelinux systemd[1]: Failed to mount /mnt/asusc.
feb 21 20:04:04 guillelinux systemd[1]: mnt-asusc.mount: Unit entered failed state.

Trying to narrow the problem, I have seen that If I disable the last line in fstab, everything goes back to normal... but the previous one looks the same to me, and it does not fail.

Now, I suppose that this bug may be closed, and maybe I should open another one if I can't find what's the real problem...

I'm sorry if my english is not very clear.
Comment by Dave Reisner (falconindy) - Monday, 29 February 2016, 11:56 GMT
If you want automounts to nicely fail when they aren't available, I think you need to include x-systemd.device-timeout=N in your fstab, where N is some value greater than 0. I don't know what changed in systemd 229, or why libreoffice tries to enumerates these mountpoints on startup.
Comment by Guillermo Leira (gleira) - Thursday, 03 March 2016, 12:23 GMT
I have tried that, but it makes no difference.

The problem is not only with libreoffice (it tries to find the documents in the recent files list), even df fails.
I have to stop all the mount and automount systemctl units:

systemctl stop mnt-nas.automount
systemctl stop mnt-nas.mount
etc.

and then everything goes back to normal.
Comment by Dave Reisner (falconindy) - Sunday, 12 June 2016, 19:44 GMT
Is this still a problem with v230?

Loading...