The daily logrotate starves the system resources
mo-gr opened this issue · 6 comments
A while ago, the hourly log rotate was re-niced. But the daily log rotate is still running at high priority.
This is the output from a recent taupage system:
$ diff /etc/cron.daily/logrotate /etc/cron.hourly/logrotate
14c14
< /usr/sbin/logrotate /etc/logrotate.conf
---
> nice -n 19 ionice -c3 /usr/sbin/logrotate /etc/logrotate.confThis also opens the question, why the daily and the hourly logrotate to exactly the same, just at different nice-levels. Maybe it's worth considering to delete the daily log rotate?
Had a discussion offline with @szuecs and tested with the latest taupage on tailor with this fix. Still seeing the latency spikes during the same time as before.
These spikes are ~every 10 minutes, for sure not caused by logrotate daily. Can you create a new issue (and close this one) or change the issue accordingly such that the subject reflects the topic?
Thanks!
Manually changed the graph timeline to show only the impact caused by the rotation. Check the full graph for 6 hours
I agree, that the daily log rotate is probably not the cause. So I'm happy to close this. But I still believe something in the daily cron job is causing this.