50% CPU usage bursts on Whonix-Gateway

I am experiencing “bursts” of Whonix Gateway using 50% CPU then going back to normal 3-4 times in a row. Exactly 50%. I have not determined a pattern yet. This is new behavior, I have not experienced it until 3 days ago. It happens infrequently and quickly for 2-3 seconds which makes it hard to launch a system monitor.

  • I have made no changes to Whonix Gateway
  • Nothing of interesting in system journal
  • sdwdate did not run
  • normal amount of open Tor circuits
  • Tor Browser was not being used at the time it happen(s)

Any recommendations to debugging this or finding a root cause? I will try to launch a system monitor, and check if Tor traffic is being sent/what amount next.

Platform: Qubes-Whonix

1 Like
1 Like

FWIW I’ve also seen this, though not at 50%, more like ~30 or so percent…didn’t think much of it so far but next time I see it I’ll try to remember checking the logs.

1 Like

I caught it. dnf-3.anondist- is what is doing it, full name cut off in top.

I changed my configuration since this post, and now it uses 25% exactly. I think it’s utilizing 1 of my cores to 100% when it’s running.

So this is Qubes-Whonix specific. Mystery solved, and a possible bug found.

2 Likes

Happening when doing dom0 updates? Using sys-whonix as UpdateVM?

Probably a bug report against Qubes needed.

1 Like

Can confirm; caught it in top as well now, it’s dnf-3.anondist-, which causes these little bursts of CPU activity for a few seconds, going up to as much as ~50% in top. After a few minutes (two or three bursts later) it seems to stop, though, and the process has terminated.

So maybe it’s just a dnf wrapper checking for updates?

Wasting time by posting here and not at Qubes.