Systemd troubles: incompatibility with journald ?
Hello list. I'm unable to run syslog-ng, whatever the version (I tried 3.3.6 and 3.4.1), on a recent system using systemd (mageia cauldron 64 bits, systemd 195). As soon as I'm launching syslog-ng, systemd and systemd-journal cpu usages reach 80/95%, and systemd fill dmesg log with this kind of message: [ 5962.484250] systemd[1]: syslog-ng.service start request repeated too quickly, refusing to start I tried to disable journald, but i didn't suceed. I also found indication of using the journald socket ( /run/systemd/journal/syslog) instead of /dev/log in syslog-ng configuration, but it doesn't change much. Also, the fedora package for syslog-ng doesn't contains anything special, either in default configuration, or in systemd service unit (excepted a patch to force pid file location). So, I'm a bit lost here... There is a user bug report on this issue: https://bugs.mageia.org/show_bug.cgi?id=8944 -- BOFH excuse #223: The lines are all busy (busied out, that is -- why let them in to begin with?).
One thing that occured to me, can you change the config to use the system() source instead of /dev/log? that should 'magically' do the right thing. ----- Original message -----
Hello list.
I'm unable to run syslog-ng, whatever the version (I tried 3.3.6 and 3.4.1), on a recent system using systemd (mageia cauldron 64 bits, systemd 195). As soon as I'm launching syslog-ng, systemd and systemd-journal cpu usages reach 80/95%, and systemd fill dmesg log with this kind of message: [ 5962.484250] systemd[1]: syslog-ng.service start request repeated too quickly, refusing to start
I tried to disable journald, but i didn't suceed. I also found indication of using the journald socket ( /run/systemd/journal/syslog) instead of /dev/log in syslog-ng configuration, but it doesn't change much. Also, the fedora package for syslog-ng doesn't contains anything special, either in default configuration, or in systemd service unit (excepted a patch to force pid file location). So, I'm a bit lost here...
There is a user bug report on this issue: https://bugs.mageia.org/show_bug.cgi?id=8944
-- BOFH excuse #223:
The lines are all busy (busied out, that is -- why let them in to begin with?). ______________________________________________________________________________ Member info: https://lists.balabit.hu/mailman/listinfo/syslog-ng Documentation: http://www.balabit.com/support/documentation/?product=syslog-ng FAQ: http://www.balabit.com/wiki/syslog-ng-faq
Guillaume Rousse <guillomovitch@gmail.com> writes:
Hello list.
I'm unable to run syslog-ng, whatever the version (I tried 3.3.6 and 3.4.1), on a recent system using systemd (mageia cauldron 64 bits, systemd 195). As soon as I'm launching syslog-ng, systemd and systemd-journal cpu usages reach 80/95%, and systemd fill dmesg log with this kind of message: [ 5962.484250] systemd[1]: syslog-ng.service start request repeated too quickly, refusing to start
As it was found in the bug report too, the problem was due to the missing /var/syslog-ng directory. A similar problem was also recently reported in Debian: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=699942 The Debian report has the better fix, I believe. But indeed, this is a packaging error, but one that should be documented, and one that can be worked around with a change to the syslog-ng unit file. I'll probably update the syslog-ng unit file, and/or write some documentation regarding syslog-ng and systemd, aimed at packagers. -- |8]
Le 08/02/2013 12:57, Gergely Nagy a écrit :
Guillaume Rousse <guillomovitch@gmail.com> writes:
Hello list.
I'm unable to run syslog-ng, whatever the version (I tried 3.3.6 and 3.4.1), on a recent system using systemd (mageia cauldron 64 bits, systemd 195). As soon as I'm launching syslog-ng, systemd and systemd-journal cpu usages reach 80/95%, and systemd fill dmesg log with this kind of message: [ 5962.484250] systemd[1]: syslog-ng.service start request repeated too quickly, refusing to start
As it was found in the bug report too, the problem was due to the missing /var/syslog-ng directory. A similar problem was also recently reported in Debian: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=699942
The Debian report has the better fix, I believe. But indeed, this is a packaging error, but one that should be documented, and one that can be worked around with a change to the syslog-ng unit file.
I'll probably update the syslog-ng unit file, and/or write some documentation regarding syslog-ng and systemd, aimed at packagers. There was a packaging error, indeed, but I'm not sure it was the only cause of the problem here, as I was able to reproduce the issue even with a correct state directory. I guess any start failure will trigger this issue, given systemd socket activation mechanism. -- BOFH excuse #370:
Virus due to computers having unsafe sex.
Guillaume Rousse <guillomovitch@gmail.com> writes:
There was a packaging error, indeed, but I'm not sure it was the only cause of the problem here, as I was able to reproduce the issue even with a correct state directory. I guess any start failure will trigger this issue, given systemd socket activation mechanism.
Yep, any start failiure will trigger it, indeed. There's not much I can do about that on the syslog-ng side I think. If there is, do let me know, though :) Come to think of it.. supporting sd_notify() might be a solution here, but I have to read up on what that exactly buys us. -- |8]
please note that the persist file should be persisted accross reboots. but you probably know this anyway. ----- Original message -----
Guillaume Rousse <guillomovitch@gmail.com> writes:
Hello list.
I'm unable to run syslog-ng, whatever the version (I tried 3.3.6 and 3.4.1), on a recent system using systemd (mageia cauldron 64 bits, systemd 195). As soon as I'm launching syslog-ng, systemd and systemd-journal cpu usages reach 80/95%, and systemd fill dmesg log with this kind of message: [ 5962.484250] systemd[1]: syslog-ng.service start request repeated too quickly, refusing to start
As it was found in the bug report too, the problem was due to the missing /var/syslog-ng directory. A similar problem was also recently reported in Debian: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=699942
The Debian report has the better fix, I believe. But indeed, this is a packaging error, but one that should be documented, and one that can be worked around with a change to the syslog-ng unit file.
I'll probably update the syslog-ng unit file, and/or write some documentation regarding syslog-ng and systemd, aimed at packagers.
-- |8]
______________________________________________________________________________ Member info: https://lists.balabit.hu/mailman/listinfo/syslog-ng Documentation: http://www.balabit.com/support/documentation/?product=syslog-ng FAQ: http://www.balabit.com/wiki/syslog-ng-faq
participants (3)
-
Balazs Scheidler
-
Gergely Nagy
-
Guillaume Rousse