On Fr, 11.09.20 09:39, Belisko Marek (marek.belisko@xxxxxxxxx) wrote: > Hi, Sorry for the late reply. > and it blocks forever. I was looking into a similar issue: > https://github.com/systemd/systemd/pull/1848 but this part is already > included in the systemd we're using (v234) and it's about persistent > logging. Any ideas on what we can check or what can cause this > behavior? This might simply be an event loop starvation issue, i.e. the sync logic runs at "idle" priority, i.e. if there's no other event to process anymore. However, if you keep bombarding the journal with log events, then this means the sync logic can be starvaed out because there's always something else to do. Note that you are running a more than 3 year old systemd version. Since then some work got merged into journald that aggressively caches process metadata so that it doesn't have to be reacquired on each log event. This means the work that needs to be done per log event is drastically reduced, thus making the problem less of an issue. That said, even then the starvation issue might still exist. We recently merged some feature work to the event loops we use that allows to address certain forms of starvation. It's not used for journald yet. Please file a bug on github about this if current journald versions still show this behaviour for you, and we can look in making use of this to make the sync stuff eventually complete even if journald is under load. Lennart -- Lennart Poettering, Berlin _______________________________________________ systemd-devel mailing list systemd-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/systemd-devel