Hi, I am running a relay server on a machine with 2GB ram. It has been running well and relayed around 7TB since the last restart. I notice that the relay server consumes over 1GB ram, over half of the system memory. Is this a bug or a feature, and can I decrease it somehow?
Is it virtual memory (VSZ) or RSS?
Output from top or something like that would help.
That’s unexpected.
Ok this is quite a lot. It seems to be spawning an awful of lot of threads which is probably the cause. I know go spawns a thread per blocking io operation, so I suspect we are not cleaning up some io operations somewhere.
I checked your status page and it seems there are 10k routines, so we are leaking them.
If output from the relay server is saved somewhere, can you send it a SIGQUIT signal (that will make it exit and dump stacktraces) and open a github issue with the stacktraces?
If the output is not stored, then restart it with output redirected, wait a few days and do the SIGQUIT trick?
What do you mean by the output? I save stdout and stderr in a log file. Does this mean, the output is saved?
Yeah. So if you send it a SIGQUIT signal, and then open a github issue with the stacmtraces that get dumped in the log, that’d be great.
ok, I opened the issue:
Thanks for that. I’ll get it fixed.
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.