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
STEP: tearing down the test environment
STEP: bootstrapping test environment
1.65058046721266e+09 DEBUG controller-runtime.test-env using existing cluster
1.650580467212675e+09 DEBUG controller-runtime.test-env automatically acquiring client configuration
Failure [0.008 seconds]
[BeforeSuite] BeforeSuite
/Users/carolynvs/src/kubernetes-plugins/tests/integration/operator/ginkgo/suite_test.go:42
Unexpected error:
<*errors.errorString | 0x14000118930>: {
s: "unable to grab random port for serving webhooks on: lstat /Users/carolynvs/Library/Caches/kubebuilder-envtest/./port-63025: no such file or directory",
}
unable to grab random port for serving webhooks on: lstat /Users/carolynvs/Library/Caches/kubebuilder-envtest/./port-63025: no such file or directory
occurred
/Users/carolynvs/src/kubernetes-plugins/tests/integration/operator/ginkgo/suite_test.go:51
It's always some % of the parallel tests that fail for me. When they do fail, I can see from the log that they all tried to use the same random port number. I'm not sure if that's the problem, or perhaps we need to call envtest differently to run in parallel safely?
When I force it to run just once, with -p -nodes 1, then the tests pass. On the CI server, it has only one vCPU so I think it's not actually running in parallel. On my machine there are definitely 4 available cores.
@sgettys@bdegeeter Can you verify that this works on your machine (with multiple cores)?
The text was updated successfully, but these errors were encountered:
When I run
make test
locally it always fails withIt's always some % of the parallel tests that fail for me. When they do fail, I can see from the log that they all tried to use the same random port number. I'm not sure if that's the problem, or perhaps we need to call envtest differently to run in parallel safely?
When I force it to run just once, with
-p -nodes 1
, then the tests pass. On the CI server, it has only one vCPU so I think it's not actually running in parallel. On my machine there are definitely 4 available cores.@sgettys @bdegeeter Can you verify that this works on your machine (with multiple cores)?
The text was updated successfully, but these errors were encountered: