Transcoder Using All System Memory

Plex Media Server
,
1

Server Version#: 1.42.2.10156

Currently running the official docker container on unRAID 7.2.2 (kernel 6.12.54). Over the past week or so I’ve had an issue where the server would run out of memory and kill a VM, and at one point it even led to a complete lockup of the server requiring a hard restart.

After letting the server run without the VM, I discovered the culprit was the Plex Transcoder, which at one point ate a whopping 53GB of RAM.

Here is an excerpt from unRAID’s syslog showing that:

Out of memory: Killed process 4127540 (Plex Transcoder) total-vm:55839484kB, anon-rss:55779032kB, file-rss:364kB, shmem-rss:0kB, UID:99 pgtables:109296kB oom_score_adj:0

I have temporarily alleviated the problem by enforcing an 8GB memory limit on the docker container, but last night the transcoder hit that limit 4 times within 10 minutes and was subsequently killed off.

Of note, these errors have all occurred during the maintenance window. Forcing transcoding even on some of my 4K content has not led to the transcoder using an extravagant amount of RAM or crashing.

I’ve tried looking through the logs myself at around each of the times the process was killed off, but I’m not well-versed enough in Plex’s logging to know what to look for.

Plex Media Server Logs_2025-11-25_09-06-18.zip (1.7 MB)

2 
Nov 25, 2025 02:06:20.119 [23063242042168] ERROR - SQLITE3:0x80000001, 11, database corruption at line 68825 of [a29f994989]
Nov 25, 2025 02:06:20.119 [23063242042168] ERROR - SQLITE3:0x80000001, 11, database corruption at line 68975 of [a29f994989]
Nov 25, 2025 02:06:20.119 [23063242042168] ERROR - SQLITE3:0x80000001, 11, statement aborts at 10: [SELECT bl.`id`,bl.`blob` FROM blobs bl WHERE (bl.`linked_type`=:C1 AND bl.`linked_id`=:C2 AND bl.`blob_type`=:C3)] database disk image is malformed

Unsure if this has to do with the memory issue, but the log files show database corruption.

Try DBRepair for Plex Media Server to resolve.

Start with the AUTO option. If unable to repair, try REPLACE to check and load a backup database.

The README has information on how to use with Docker.

1 Like
3

Ran through it and it says it was able to repair the Blobs DB, so I guess I’ll keep an eye on it and see if there’s OOM reports in the morning.

EDIT: Well, no out of memory reports in the syslog so I’m going to go ahead and call this one solved. Appreciate the help!