Server 3.12.5 Crash

[anprdev]

Hi all
We have talk about this long time ago.
My Server on my Raspberry PI crash somehow. This can happening in 1 day or weeks. I can not get my finger on it.
Blues/Rock Server has default 5 clients on it all version 3.12.5
The same issue was also on older versions.
I start and log the Server with a script
nohup bash -c '
jamulus_server -s -T -F -u 20 --jsonrpcport 25112
--discononquit
--welcomemessage /home/pi5/JamulusStartup/Jamulus-Welcome-Short.html
--jsonrpcsecretfile /home/pi5/JamulusStartup..........secret
2>&1 | cronolog /home/pi5/JamulusStartup/WebServerPlayer/JamServerLOG/jamulus_%d-%m-%Y.log
' >/dev/null 2>&1 &
Every day I get a new log file.
The bad thing is that I don't get a error when the Server crash.
All clients are disconnected when the crash is happening and the server is frozen.

My Question is there a other option to get error messages If the server is sending them with a crash.

[ann0see]

This seems bad. Please open an issue for this. We've been hunting down potential crash sites over the last few months but might not have found all of them yet.

[ann0see]

Please also state which JSON-RPC methods you use.

[anprdev]

JSON-RPC is only enable and the only thing I can do on the server with this is set the server Private but most of the time I don't use it and do it at the server itself set directory to None.
So the Server only is waiting for a command.
I also get the idea that the crash is happening when somebody is log in to the server. Maybe some wrong package when a client connect but I can not find this and can not try too reconstruct this crash.
The Server can run for weeks or the crash can happen twice a day.
Maybe clients with old versions can be a issue.
The issue is happening also with the older version like 3.11
The connection to the server is only when the command is send. After sending the command to set private the server is disconnect same for public.
So this can not be a problem.

[pljones]

The issue is happening also with the older version like 3.11

Is there any version where it doesn't happen?

[anprdev]

I don't can remember what version I started with. but the issue is started with 3.11maybe something different there.

[mcfnord]

🤖 AI: The launcher discards the one place a crash gets reported: bash prints Segmentation fault (core dumped) plus the exit status on its own stderr, and >/dev/null 2>&1 throws that away.

Tested on a Pi 4 running Debian 13: the kernel log stays silent, but coredumpctl list shows the crash and coredumpctl info jamulus_server prints a backtrace (needs the systemd-coredump package).

Two changes to the script would answer ann0see's question at the next event: ulimit -c unlimited before the server line, and >>/home/pi5/JamulusStartup/launcher.log 2>&1 in place of >/dev/null 2>&1.

The daily log already records every join as <time>, <ip>, connected (<n>), with or without -l (AddNewConnection prints to the console unconditionally), so its last lines before the event show whether a join preceded it.

If the process is still alive afterwards, attaching gdb with sudo gdb -p $(pgrep jamulus_server) -batch -ex 'thread apply all bt' shows where every thread is stuck. Please post the coredumpctl info output in the issue.

[mcfnord]

Are you using the same sdcard for the O.S. ? I've seen a failing sdcard cause this kind of failure.

[anprdev]

No SD-Card all on Samsung SSD