Killed by OOM after couple of days recording rtsp to hdd #4674
Description
Which version are you using?
1.12.3
Which operating system are you using?
Linux amd64 standard
Describe how to replicate the issue
MediaMTX is started as systemd service (via simple .sh only starting binary), config file is default without any paths and API enabled
10 paths are added via api with RTSP over TCP enabled
2 paths are completely dead (not connected to network)
2 paths are not sending any data (mediamtx reports paths are ready but received data is not growing)
for all paths storing to hdd is enabled recordFormat = "fmp4", recordPartDuration = "1s", recordSegmentDuration = "10s", records are not deleted by mediamtx (recordDeleteAfter = "0s") but deleted by another software by timer
Once a second mediamtx is polled via api with /v3/paths/list?itemsPerPage=100 request
First time everything is ok and ram usage is low, but after couple of days mediamtx is suddenly killed with OOM
Tasks state (memory values in pages):
[ pid ] uid tgid total_vm rss rss_anon rss_file rss_shmem pgtables_bytes swapents oom_score_adj name
[ 21754] 0 21754 6784615 3308362 3308330 32 0 30695424 498464 0 mediamtx
Out of memory: Killed process 21754 (mediamtx) total-vm:27138460kB, anon-rss:13233320kB, file-rss:128kB, shmem-rss:0kB, UID:0 pgtables:299>
Logs shows nothing in mediamtx ouput (just complains about 2 not connected publishers)
No one is reading streams from mediamtx when ram leak happens
How can I get better memory usage profile for mediamtx to understand what is the reason of OOM?
Server logs
No response
Network dump
No response