Syncthing stops working after some time (process unresponsive and using almost no RAM)

This issue doesn’t seem to happen very often, but we’ve experienced it at least a few times. Only one computer is affected by it. Basically, Syncthing appears to just “die” in background, becoming completely unresponsive with its RAM usage dropping to almost nothing.

This is on Windows 10 x64 with Syncthing 1.20.4. The situation in the Task Manager looks as below (in Korean, but I hope the numbers are clear — no CPU usage, and RAM usage counted in KB).

Have you got any ideas what the culprit could be here? Has anyone else ever encountered a similar issue?

For the record, there’re no panic logs or anything in the Syncthing logs really. It just seems to stop working at some point for no apparent reason.

Enabling the built in profiler and taking a goroutine trace will show what it’s doing.

This happened again today. I’d set up the profiler beforehand, but once the RAM dropped to almost nothing, http://localhost:9090/debug/pprof wasn’t responding either. I then checked the Task Manager and the RAM was basically 100% full. I then found a rogue program that was using almost half of the memory and killed it, which dropped the RAM usage to ~30%.

What’s interesting is what took place next. Basically, Syncthing came back to life on its own. It wasn’t a continuation of the previous run, as the uptime had been reset, but it seems that it did somehow manage to restart itself. I don’t think it actually crashed, as there were no panic logs saved in the home folder. Not really sure what’s going on there exactly, but it appears to be the OS forcefully dropping the RAM usage of other programs to a minimum, including Syncthing, perhaps to prevent crashing due to being out of memory.

Just for the record, the rogue application was https://github.com/rustdesk/rustdesk (a remote desktop tool).

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.