Ubuntu
systemd package

  1. Bug #1829829
  2. Comment #18

Comment 18 for bug 1829829

Revision history for this message
Dan Streetman (ddstreet) wrote :

> right now the upstream test timeouts on i386

ah, ok - so I did look into this, or at least some failures really similar to this, a while back, for bug 1831468. As far as I could tell, the problem in those cases was that the test was starting a service and waiting for notification of its completion; but for some reason, dbus seemed to be crashing/dying/exiting, and the notification of the completed service never actually reached the test suite, so it just waited forever until timeout. I do see in the logs for the test you referenced:

Unable to add match type='signal',sender='org.freedesktop.DBus',path='/org/freedesktop/DBus',interface='org.freedesktop.DBus',member='NameOwnerChanged',arg0=':1.7', failing connection: Method call timed out
Bus bus-api-system: changing state RUNNING → CLOSING
Unable to add match type='signal',sender='org.freedesktop.DBus',path='/org/freedesktop/DBus',interface='org.freedesktop.DBus',member='NameOwnerChanged',arg0=':1.7', failing connection: Connection terminated
Bus bus-api-system: changing state CLOSING → CLOSED
systemd-rfkill.socket: Changed dead -> listening
systemd-rfkill.socket: Job 364 systemd-rfkill.socket/start finished, result=done
[ OK ] Listening on Load/Save RF …itch Status /dev/rfkill Watch.
Bus bus-api-system: changing state UNSET → OPENING
Bus bus-api-system: changing state OPENING → AUTHENTICATING

that's quite similar to what i saw before. I never had a chance to get farther than that, but I can try to look more into it soon. Let me know if that sounds familiar or if you have any debugging suggestions.

> Plus, in https://github.com/systemd/systemd/pull/12861#issuecomment-506556062 the
> test is completely broken on Ubuntu CI and I have no idea why.

ugh, yeah TEST-15 and TEST-22 aren't giving any clues at all in the log to why they failed.

> As far as I can tell, it has been abandoned since then. If you're ready to keep
> the list up to date by visiting PRs where Ubuntu CI fails I'm of course all for it.

Let's add TEST-15 and TEST-22 for now, and I'll see if I can both keep up with other flaky tests, and also try to figure out why they're being flaky in ubuntu-ci.