There are a number of issues posted to the jellyfin GitHub about this, or something similar - theories are that it is related to a possible memory leak somewhere, or that the library scans are starting too many parallel processes as the defaults are one per processor which is too many for a quad core Pi system. I have recently started seeing this on a 4GB Pi4 since upgrading to jellyfin 10.10.3 (latest current). Same symptoms, can't contact server though still pingable. Some relevant log entries are:
A recommended workaround is to reduce the scan tasks limit to 1 (General/Performance admin tab), see att. Haven't seen any more OOM events or lockups since doing this but YMMV.
HTH
Chris K
Code:
(zzz)@tvs:~ $ journalctl | grep "Out of memory"Nov 18 00:09:26 tvs kernel: Out of memory: Killed process 1175 (jellyfin) total-vm:274498096kB, anon-rss:499772kB, file-rss:200kB, shmem-rss:21504kB, UID:1000 pgtables:2604kB oom_score_adj:0Nov 18 12:34:28 tvs kernel: Out of memory: Killed process 171793 (jellyfin) total-vm:274442324kB, anon-rss:244704kB, file-rss:128kB, shmem-rss:7804kB, UID:1000 pgtables:1712kB oom_score_adj:0Nov 18 20:53:41 tvs kernel: Out of memory: Killed process 201184 (jellyfin) total-vm:273888812kB, anon-rss:6704kB, file-rss:360kB, shmem-rss:3200kB, UID:1000 pgtables:1096kB oom_score_adj:0(zzz)@tvs:~ $HTH
Chris K
Statistics: Posted by cjketle — Thu Nov 21, 2024 2:51 pm