Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.user > #241938 > unrolled thread

Occasional failure of X to start

Back to article view | Back to linux.debian.user

Contents

  Occasional failure of X to start Richmond <richmond@criptext.com> - 2021-11-07 21:20 +0100
    Re: Occasional failure of X to start "Thomas Schmitt" <scdbackup@gmx.net> - 2021-11-07 22:00 +0100
      Re: Occasional failure of X to start Richmond <richmond@criptext.com> - 2021-11-07 23:20 +0100
        Re: Occasional failure of X to start "Thomas Schmitt" <scdbackup@gmx.net> - 2021-11-08 10:30 +0100

#241938 — Occasional failure of X to start

I have a laptop with an nvidia graphics card, and a TV connected to the
TV out (svideo) socket. Sometimes X fails to start. All I can see in the
journal is:

Nov 07 19:38:37 systemd[1]: xdm.service: Main process exited,
code=killed, status=12/USR2 Nov 07 19:38:37 systemd[1]: xdm.service:
Failed with result 'signal'.

Where should I look for a more detailed error?

If I log in on the console and use startx then X starts.

i nvidia-legacy-340xx-driver - NVIDIA metapackage (340xx legacy version)

Linux 4.19.0-18-amd64 #1 SMP Debian 4.19.208-1 (2021-09-29) x86_64
GNU/Linux

I am using Debian 10 because this driver doesn't like kernel 5.

I am using xdm probably because I had much trouble getting things to
work with the additional monitor, so although I installed KDE I am now
using icewm with xdm.

[toc] | [next] | [standalone]

#241939

Hi,

Richmond wrote:
> Nov 07 19:38:37 systemd[1]: xdm.service: Main process exited,
> code=killed, status=12/USR2 Nov 07 19:38:37 systemd[1]: xdm.service:
> Failed with result 'signal'.

Googling "xdm USR2" brings me to
  https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=948346
which shows the same message.

The bug seems to be still open and tagged with "help":
  https://www.debian.org/Bugs/Developer#tags
  "help
   The maintainer is requesting help with dealing with this bug.
   Either the maintainer does not have the skills necessary to fix this
   bug and desires collaboration, or is overloaded and wants to delegate
   this task. [...]"

I understand that logrotate sends signal USR2 to xdm to tell it that
it shall close and reopen the log file, because the old one was renamed
and a new current log file was created.
Introduced possibly because of
  https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=285871

It would be interesting to learn whether after such a start failure
on your machine the file
  /var/log/xdm.log
shows indications that it was created newly.


Have a nice day :)

Thomas

[toc] | [prev] | [next] | [standalone]

#241941

"Thomas Schmitt" <scdbackup@gmx.net> writes:

> Hi,
>
> Richmond wrote:
>> Nov 07 19:38:37 systemd[1]: xdm.service: Main process exited,
>> code=killed, status=12/USR2 Nov 07 19:38:37 systemd[1]: xdm.service:
>> Failed with result 'signal'.
>
> Googling "xdm USR2" brings me to
>   https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=948346
> which shows the same message.

OK so I should have thought of doing that.

> The bug seems to be still open and tagged with "help":
>   https://www.debian.org/Bugs/Developer#tags
>   "help
>    The maintainer is requesting help with dealing with this bug.
>    Either the maintainer does not have the skills necessary to fix this
>    bug and desires collaboration, or is overloaded and wants to delegate
>    this task. [...]"
>
> I understand that logrotate sends signal USR2 to xdm to tell it that
> it shall close and reopen the log file, because the old one was renamed
> and a new current log file was created.
> Introduced possibly because of
>   https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=285871
>
> It would be interesting to learn whether after such a start failure
> on your machine the file
>   /var/log/xdm.log
> shows indications that it was created newly.
>
>

-rw-r--r-- 1 root              root    6656 Nov  7 19:38 xdm.log.1
-rw-r--r-- 1 root              root       0 Nov  7 19:38 xdm.log

It seems so. I hadn't noticed it was always on a Sunday but could be.

I should configure logrotate /etc/logrotate.d/xdm remove weekly and put
size=1G then hopefully it won't happen for a few years?

[toc] | [prev] | [next] | [standalone]

#241948

Hi,

i wrote:
> > It would be interesting to learn whether after such a start failure
> > on your machine the file
> >   /var/log/xdm.log
> > shows indications that it was created newly.

Richmond wrote:
> It seems so. I hadn't noticed it was always on a Sunday but could be.

So it looks really like bug 948346.
(My unqualified theory is that xdm writes a log message line which
triggers the log rotation before xdm has prepared for catching and
processing signal USR2. This would explain why the rotate can reliably
hit the short timespan when xdm is vulnerable.)

Consider to send a mail to 948346@bugs.debian.org saying that you
experience the problem too.
Eduard Bloch, a (former ?) Debian Developer, seems to have a workaround
but Julien Cristau (current Debian Developer of xdm ?) seems not to accept
it.
Maybe more bug confirmations can help Edi's workaround to get in.


> I should configure logrotate /etc/logrotate.d/xdm remove weekly and put
> size=1G then hopefully it won't happen for a few years?

Looks worth a try. If it works, then consider to mention it in your mail
to 948346@bugs.debian.org .


Have a nice day :)

Thomas

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.user

csiph-web