Migration complete! Let's see how long this little honker runs.

@knapjack The Pi should be pretty stable. Honk on!

@petersanchez @dirk Ran into an interesting issue with the Pi. The backend is failing about once a day. Exactly once a day, I asked myself? Yes, it appears so.

When I migrated over, I thought what the heck, let's log to syslog. I installed socklog, left the defaults, voila.

It looks like when the daily log rotation kicks in, the backend crashes. If I try a simple restart of honk the backend doesn't like the socket. A full reboot does the trick.

So, in experimentation mode, now logging to file again. Start the clock.

@knapjack the backend may have been killed to memory pressure. but it's weird it wouldn't reconnect. the restarted process should clean up the old socket.

@tedu @petersanchez @dirk Well, even not logged to syslog, seems to be related:
2023/11/18 02:58:49 lost the backend: %!s(<nil>)
and:
-rw-r--r-- 1 root socklog 44717 Nov 18 02:58 current
Going to try disabling socklog and see what happens.

@knapjack I keep forgetting to add the err to that log.

@tedu @petersanchez @dirk So far, stopping socklog seems to have done the trick. Going to give it 24 hours, but joined the mailing list already. Should be interesting to troubleshoot.

@petersanchez @dirk @tedu Oooh, the plot thickens. Backend crashed anyway, but no problems restarting it, no reboot necessary.

Now I'm wondering if it's a musl issue, or if someone I follow is just posting something crazy daily. 😂

@knapjack @dirk @tedu Is this running on Alpine?

@petersanchez @dirk @tedu Void, so close.

@petersanchez @dirk @tedu Alright, winds gusting to 70 miles per hour, plenty of power outages, so I took it as an omen to learn the Way of Void this evening and created runit scripts for honk so now it autostarts after reboot, and auto-restarts if the backend dies. Not perfect, but I'm well-versed in the way of duct tape.