High CPU Use on Server When Desktop is Not Connected

I’ve read the FAQ and understand why Synching can use a lot of CPU. Still, I’m a bit puzzled by the current situation, and thought it was worth mentioning.

Background:

I have an i7 Dell Chromebox running Ubuntu Server with Syncthing. I also have Syncthing running on a Raspberry Pi server, a smartphone, a desktop, and occassionally on two laptops.

The folders synced with all devices except the desktop are about 2Gb. The i7 server and the desktop are about 2.5Tb. Generally speaking everything runs smooth, and I’m happy with Syncthing.

Issue:

When I turn off my desktop, my i7 server starts acting weird. The fan kicks off very loudly every minute for about ten to fifteen seconds, then quiets down.

If I SSH via my mobile phone and run top, I see that it’s Syncthing that’s likely causing this. Just as the fan comes on, Synching moves to top of the process chart, using over 100% of CPU, and a considerable amount of memory. After some ten seconds or so, it moves off the top of the process table.

There are no other major programs running on this server. The only other device with Syncthing running on it at this time is the Raspberry Pi server.

What really confuses me is that this only happens when the desktop is turned off. When the desktop is on and Syncthing has to deal withh terabytes of data, everything is quiet. Once the desktop is off, there’s blazing CPU fan noise every minute.

Questions:

Is this expected behavior? Is there anything that I can do on my end, besides killing SSyncthing on whenit’s not needed?

Lastly, sorry for typos – writing from my mobile phone.

Given the once a minute cadence I’d say “scanning” but there should be no difference between if it’s connected to something or not. Perhaps run with -verbose and see if you can correlate to what’s happening at that time.

Thanks.

Ran verbose via SSH from my phone.

This is what seems to be happening:

Folder is idle and scanning for about four or five times.

Then “folder isn’t making any progress, pausing puller for one minute”.

After which a ton of errors get dumped, all seem to be “Err:parent is not a directory.”

This definitely happens around the time that the fan goes crazy. Just not sure if it’s due to the errors. It may be starting before them.

I’ll try to pull some more details tomorrow.

Thanks again and hope this is somewhat useful.

Just wanted to clarify the error that I am getting.

Folder “folder-name” is now scanning Folder “folder-name” is now idle

This repeats four times over the course of one minute.

On the fifth time, I get:

FolderErrors events.Event {SubscriptionID:68, GlobalID:68, Time:time.Time{sec6361900112990, nsec:950844675, loc:(*time.Location)(xff26e0)}, Type:4194304, Data:map[string]interface {}{“folder”:“folder-name”,“errors”:[]model.fileError[model.fileErorr{path:“a/path/to/a/file”, Err:“parent is not a directory”}, . . .

This goes on, with different filenames, but the same error. It’s a very long list, filling the terminal screen on the mobile phone maybe a dozen times over.

(I had to copy the above from screenshots, apologies for any typos.)

At the end of the list, the scanning/idle process starts again, followed by the errors again.

Again, I’m not sure if this is the cause of the cpu/fan issue or not. But it’s the only lead that comes up at this point. Hope this is helpful.

It’s not, it’s an issue we are aware off, and are discussing a fix.

Thanks for letting me know – and for all your work. Just donated.

2 Likes

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