Random restarts with no errors in log #4022
Replies: 1 comment 5 replies
-
|
Hi @cbnathanael, |
Beta Was this translation helpful? Give feedback.
-
|
The journal doesn't have anything of interest. I've started the containers back up to see if there's anything in them, but all I've ever seen is the same as what's in the liquidsoap log. Maybe icecast is different. |
Beta Was this translation helpful? Give feedback.
-
|
I can't tell the reason for the liquidsoap restart because there are no errors in it. Somewhere near that moment there should be at least something in the journal. I would recommend using ffmpeg decoder and encoder instead of ogg. # settings.decoder.priorities.ogg := 20
# settings.decoder.file_extensions.ogg := ["ogv", "oga", "ogx", "ogg", "opus"]encoder = %ffmpeg(format="ogg", %audio(codec="flac")) |
Beta Was this translation helpful? Give feedback.
-
|
I've switched to using ffmpeg as you recommended. Just waiting for it to crash, now, and I'll double-check the system journal (both host and inside the containers), and see if there's anything unique in the docker logs as well. Thanks so much for the help! |
Beta Was this translation helpful? Give feedback.
-
|
Follow-up, @vitoyucepi -- I initially just did a copy/paste of your recommendation to switch to ffmpeg, which did nothing but add an "unused variable" message when LS started. But, setting my Icecast output to use the encoder ended up with zero crashes in 48 hours. Which, I think solved the problem! Changed |
Beta Was this translation helpful? Give feedback.
-
|
So the next step is to get the core dump and fix the segfault, if there was one. The whole process may look similar to #1996. But I need to see the syslog message first. |
Beta Was this translation helpful? Give feedback.
-
I'm running Liquidsoap 2.2.5 with Icecast 2.2.4 in docker (separate containers, combined compose.yml) and it appears that LS is randomly restarting itself when the request queue is empty and starts playing the single fallback file. However, it's not every time it plays the fallback. Sometimes it runs for a couple hours before experiencing an issue, other times it's 20 minutes. I'm unable to find any common factors other than when the stream switches to the fallback.
Initially, I thought it may be an issue with the system it was on because it was an already busy server with a lot of things running. Currently, it's on a new n100 mini pc () with little else running and it behaves identitcally.
The request queue is handled by a separate application via some harbor api endpoints. All of the files are .oga and ogg/flac.
Basic monitoring from the system these are all running on shows Liquidsoap at a stable 14% CPU and never topping 300MiB RAM. (The grafana graph shows where one instance dies and the other starts up, though it's interesting that there's a 5 minute "cool down" for the prior container)
Additionally, I do hear some playback pause/hiccups when connecting to icecast via VLC on my phone, but not anywhere else. This may just be a buffering/bandwidth issue though.
EDIT: SOLVED -- Switching to use ffmpeg encoder in the icecast output resolved the unexplained crashing
(disclaimer: yes, this is streaming copyrighted material. no, it is not streaming outside of my home)
I've attached full logs and config files at the end (and they contain multiple instances of LS restarting), but these are what I think are the pertinent bits:
Liquidsoap log:
Icecast Error Log
Icecast Access Log
quickstart.liq.txt
icecast-access.log
icecast-error.log
liquidsoap.log
docker-compose.yml.txt
icecast.xml.txt
Beta Was this translation helpful? Give feedback.
All reactions