On Wed, 10 Jan 2018, Marc Haber wrote:
>
> googling for the error message leads me to #857904 which has the same
> error with apache2.
I see. Definitely similar.
> This turned out that systemd.log_level=debug was set
> in /etc/default/grub. That one should also show up in /proc/cmdline.
Brilliant! That checks out (sort of).
I do have this 'debug netconsole=...' in /proc/cmdline (but no
systemd.log_level=debug though) as I'm chasing some misterious reboots
which leave no traces in the logs, using netconsole. But 'debug' is a
kernel directive, isn't it? Why should systemd inherit that?
> And
> you should get this blurb every time a systemctl call is executed, for
> example when logrotate rotates a log and kicks the daemon later.
>
> Atop is a likely culprit for that as it's restarted every night.
>
> If this is the case then my first assessment that you're running in some
> debug mode is true.
Yes, see above. And I do confirm /var/log/auth.log is full of these:
systemd-logind[952]: Got message type=signal sender=:1.1 destination=n/a
object=/org/freedesktop/systemd1/unit/atop_2e*
messages, growing rapidly (tens of thousends of messages in just one
day).
Another observation. If I run:
# systemctl restart --quiet atop
from command line, nothing comes out on stdout/stderr.
> The systemd log level can also be set in the [Manager] section of
> /etc/systemd/system.conf
This smells like a systemd bug (inheriting the kernel 'debug'
directive), long way. For now, I modified /etc/systemd/system.conf
and did:
# systemctl daemon-reload
hoping it would do what it promices:
Reload the systemd manager configuration.
But it doesn't :( 'daemon-reexec' doesn't change anything either.
Restarting the systemd-logind service doesn't help either. I'm not
jumping to restart the computer as I think all points to leading to
anything different, unless someone confirms that it will.
My opinion is that bug belongs to systemd. What do you think?
Cheers,
--
Cristian
