CI fails when other jobs are running because of time out.

[n-eiling]

See https://git.rwth-aachen.de/acs/public/villas/node/-/jobs/5510111 for details.

Relevant output:

[TIME]  pipe-loopback-zeromq (ran for more then 1m)
[SKIP]  pipe-python-protobuf: Test is broken
[SKIP]  relay: Test is broken
[PASS]  signal (ran for .100993659s)
[PASS]  test-config (ran for 7.858488891s)
Summary: 1 of 61 tests failed.
   Timedout: 1
   Skipped: 18

Maybe we can remove the time out?

[stv0g]

I guess the timeout is probably more of an indicator for a broken test?

Shouldn't we attempt to fix the test, or disable it?

[n-eiling]

The test does work if it I rerun it. So it seems to work, but is just too slow if run in parallel with everything else.
I haven't investigated thoroughly though and I don't know if a similar functionality is tested in another test.

[stv0g]

Is this still happening? I've seen it not myself yet..

[n-eiling]

Haven't seen it in a while - but that doesn't mean it fixed itself. Don't you think it's bad practice to rely on timing in the CI? The runner might be really slow sometimes.

[stv0g]

Yeah, its a valid consern.

I have added those timeout checks to avoid clogging the CI with jobs which are dead-locked.
Maybe we should raise the timeout to something like 1h?

[n-eiling]

sounds good to me

[IgnoreWarnings]

Haven't seen it in a while - but that doesn't mean it fixed itself. Don't you think it's bad practice to rely on timing in the CI? The runner might be really slow sometimes.

Is this still happening? I've seen it not myself yet..

I got the timeout yesterday, it is still there.

[IgnoreWarnings]

I get it consistently now on every single commit.

[IgnoreWarnings]

Increasing the timeout didnt help, the integrationtest seems to get stuck somehow.

[IgnoreWarnings]

I guess the timeout is probably more of an indicator for a broken test?

Shouldn't we attempt to fix the test, or disable it?

The test does work if it I rerun it. So it seems to work, but is just too slow if run in parallel with everything else. I haven't investigated thoroughly though and I don't know if a similar functionality is tested in another test.

It looks like @stv0g is right that the test is broken. I increased the timeout and the CI hangs on integration test.

[IgnoreWarnings]

"[TIME] pipe-loopback-nanomsg (ran for more then 15m)"

[IgnoreWarnings]

But it looks like before it was on another test: [TIME] pipe-loopback-zeromq (ran for more then 1m)

[stv0g]

Okay, if the tests are broken, we need to figure out what broke them as they used to work in previous commits.

Maybe git bisect could help..

[n-eiling]

I suspect this is more due to docker / strange interactions between the tests.