Closed abdulocracy closed 2 years ago
I've rolled back to 3.29 and it still crashes with the same message. This started after upgrading to 3.30 yesterday.
I restored to a backup from before the upgrade, and can recreate the breakage introduced by 3.30.
Looks like there is no memory available. Could you check memory when running lms? Is there something special you do on the interface to make it crash?
I wasn't doing anything particular, just starting to play any song. I can't reproduce the behavior right now, perhaps the server was in an OOM state when I reported the bug, however unlikely it seems. Please close, I'll reopen if I can reproduce. Thanks. :)
I'm still experiencing this issue, the server is nowhere near OOM and the container is far from its limit. I don't think it's an OOM issue.
Here's a slightly more detailed crash log.
Edit: never mind, it's the same log, for some reason it's malformed. Maybe less butchered it.
lms | 172.20.0.2 - - [2022-Jul-13 06:01:32.484] "POST /?wtd=pShiPWMX3Hgko2Jx HTTP/1.1" 200 0
lms | 172.20.0.2 - - [2022-Jul-13 06:01:32.485] "POST /?wtd=pShiPWMX3Hgko2Jx HTTP/1.1" 200 96
lms | [flac @ 0x7f768ed85680] Could not find codec parameters for stream 1 (Video: mjpeg, none): unspecified size
lms | Consider increasing the value for the 'analyzeduration' and 'probesize' options
lms | 172.20.0.2 - - [2022-Jul-13 06:01:32.711] "GET /release/mbid/?wtd=pShiPWMX3Hgko2Jx&request=resource&resource=o1rj
7an6&ver=1&trackid=45750&size=128 HTTP/1.1" 200 4165T /release/mbid/?wtd=pShiPWMX3Hgko2Jx&request=resource&resource=o1rj
7anc&ver=1&trackid=45750&bitrate=192000&format=2 HTTP/1.1" 200 4238896ht in IO context: std::bad_alloc
Ok, will try the docker version to see if I can reproduce. Here what I get so far: Looks like the RES memory usage is quite limited for now
By the way, did you set some constraint? (see https://docs.docker.com/config/containers/resource_constraints/)
I have no resource constraints on the container, so it's got plenty of room to grow when it fails.
Another question: do you play transcoded tracks?
Could you by chance run using full debug output? log-config = "*";
in lms.conf
Yes, I play transcoded tracks. Here's the full log: lms.log
It seems to always happen when the 'love' feedbacks are queued to be sent.
Ok, thanks for the valuable log file. I have found something bad in the code involved that is very likely to explain your crash. Will schedule a hotfix asap
Thank you very much for the prompt fix. :)
@abdulocracy can you update your docker image and confirm the problem is gone for you?
I've updated, will test and let you know if the issue persists.
I'm running 3.30.0 on Docker x64.
It crashes with the following logs repeatedly: