canonical / microcloud

Automated private cloud based on LXD, Ceph and OVN
https://microcloud.is
GNU Affero General Public License v3.0
260 stars 36 forks source link

Fix test suite hangs #293

Closed masnax closed 2 months ago

masnax commented 2 months ago

For the debug binaries and ZFS, I'm not sure exactly why the timeouts are necessary but without them, I pretty consistently run into issues where the action is actually performed, but the command won't let go.

masnax commented 2 months ago

@simondeziel did you ever run into issues with the places I put timeouts in?

Nevermind, it turned out to be lxc exec ${name} -- rm -f {path_to_binary}.debug getting stuck if the path did not exist. I don't know why this caused the command to hang but wrapping it in an if block fixes the problem.

simondeziel commented 2 months ago

@simondeziel did you ever run into issues with the places I put timeouts in?

No

masnax commented 2 months ago

@simondeziel did you ever run into issues with the places I put timeouts in?

No

You wouldn't have, since I removed the commits :)

The issue seems to be related to RUN_CONCURRENT=1

Seems when we detach the process and then run lxc exec -- {something}, if the command returns too quickly then it never exits. Not sure if it's just a bug with my terminal.

simondeziel commented 2 months ago

Seems when we detach the process and then run lxc exec -- {something}, if the command returns too quickly then it never exits. Not sure if it's just a bug with my terminal.

@MusicDin does that ring a bell from when you investigated/worked on a lxc exec bug a while ago?