Syncthing 1.17.0 on Ubuntu 20.04 VM systemd starting too many copies

I just set up a VPS with a provider. The instance of ubuntu os stock. I installed syncthing 1.17.0 on it and set up a single directory to test speeds from it to my home network.

It is set up with the default systemd scripts. I have enabled the service to autostart for my username (non root) on the server.

It initially works perfectly. Syarts syncing my files as expected but starts slowing down. Turns out systemd keeps launch8ng new copies of the syncthing process until physical and virtual memory is exhausted. Going on with an high system load until the server becomes completely unresponsive and the provider’s VM triggers a restart of my server.

It takes about 10 minutes for this cycle to occur give or take.

Would repeat, I have not modified any systemd startup files, except to enable the syncthing@username for my account and then start it up. Configuration is very simple. A user login for the web GUI, one sync folder. Send only, no versioning, sync oldest first.

Has anyone encountered this situation?

Any logs?

At least the output of systemctl status syncthing@username.service would be helpful.

Neglected to mention… I dumped logs from journalctl. It has nothing that seems distressing. It is attached here

log.txt (15.6 KB)

[monitor] INFO: Signal 15 received; exiting

Your service receives a SIGTERM signal.

Yes. Likely caused by the host after a few hundred processes were spawned and whatever monitoring software they have in place restarted the VM…

Are you using this version?

Yes, that is the file that was installed on my system. It is unedited.

Anything in the kernel logs?

dmesg -eT

Output of dmesg is completely empty. I have not cleared it…

Is your journalctl output filtered? Is systemd loggong anything on its own?

Reviewed the last two hours of output from journalctl. Just the same syncthing output, entries of evidence from random bots trying to ssh their way in…. No errors.

I have not set up any filtering but I don’t know if the image I installed from has anything set up.

Tentatively resolved.

On a whim, I dropped the 1.18.0-rc1 binary over top of 1.17.0 and restarted it.

I am Watching the load average consistently trend downward and it is not spawning more tasks.

Will monitor it tonight and see what happens.

It is in fact not resolved, but the problem ramps up over an hour now instead of 10ish minutes

Are you sure that your provider isn’t interfering here?

Yes, quite confident of that.

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