You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
test-cluster-disconnect-handles hanging in background on freebsd #6988
If nothing is jumping out at anyone in the code, maybe try running this test 10K times on freebsd in the stress test job and see if it fails at some point (and see if that failure causes subsequent runs to fail too). And, of course, fix whatever is causing the first failure.
Expanding scope a bit, judging from failures like https://ci.nodejs.org/job/node-test-commit-arm/3454/nodes=armv7-ubuntu1404/consoleFull it seems that some test somewhere is not cleaning up after itself reliably, even when it succeeds. I imagine it would manifest as a stray node process running and keeping the port open, resulting in the plethora of EADDRINUSE errors. I imagine this is a likely source of this issue. Not sure of an easy way to identify the problematic test, though. EDIT: Oh, actually, maybe this is the culprit test...
Activity
rvagg commentedon May 26, 2016
@nodejs/testing
Trott commentedon May 26, 2016
If nothing is jumping out at anyone in the code, maybe try running this test 10K times on freebsd in the stress test job and see if it fails at some point (and see if that failure causes subsequent runs to fail too). And, of course, fix whatever is causing the first failure.
I guess in the time it took me to type that out, I could have just kicked off the job. So here it is: https://ci.nodejs.org/job/node-stress-single-test/744/nodes=freebsd102-64/console
cjihrig commentedon May 26, 2016
The stress test came back green. Are there any processes left behind?
Trott commentedon May 26, 2016
Expanding scope a bit, judging from failures like https://ci.nodejs.org/job/node-test-commit-arm/3454/nodes=armv7-ubuntu1404/consoleFull it seems that some test somewhere is not cleaning up after itself reliably, even when it succeeds. I imagine it would manifest as a stray node process running and keeping the port open, resulting in the plethora of EADDRINUSE errors. I imagine this is a likely source of this issue. Not sure of an easy way to identify the problematic test, though. EDIT: Oh, actually, maybe this is the culprit test...
test: verify cluster worker exit
test: verify cluster worker exit
cjihrig commentedon Jun 1, 2016
@Trott, @rvagg can this one be closed? The issues with the CI from last week seem to be cleared up.
Trott commentedon Jun 1, 2016
I think it can be closed.
cjihrig commentedon Jun 1, 2016
OK. Closing. Please reopen if you think it needs to be investigated further.
test: verify cluster worker exit
test: verify cluster worker exit
test: verify cluster worker exit
test: verify cluster worker exit
test: verify cluster worker exit
test: verify cluster worker exit