Closed t0yv0 closed 7 months ago
:tropical_drink: The Update for pulumi/k8s-ci-cluster/dcb98b0ba0ac5801fe1b47555bcddeadccc49832-1687 was successful.
Name Type Operation
+ multicloud pulumi-kubernetes:ci:GkeCluster create
+ password random:index/randomPassword:RandomPassword create
+ ephemeral-ci-cluster gcp:container/cluster:Cluster create
+ primary-node-pool gcp:container/nodePool:NodePool create
+ gke pulumi:providers:kubernetes create
+ k8s-ci-cluster-dcb98b0ba0ac5801fe1b47555bcddeadccc49832-1687 pulumi:pulumi:Stack create
Looks like the test failure is an instance of https://github.com/pulumi/pulumi-gcp/issues/1387. I closed the original issue due to a missing repro, but I am re-opening.
:tropical_drink: The Destroy for pulumi/k8s-ci-cluster/dcb98b0ba0ac5801fe1b47555bcddeadccc49832-1687 was successful.
Name Type Operation
- gke pulumi:providers:kubernetes delete
- primary-node-pool gcp:container/nodePool:NodePool delete
- ephemeral-ci-cluster gcp:container/cluster:Cluster delete
- multicloud pulumi-kubernetes:ci:GkeCluster delete
- password random:index/randomPassword:RandomPassword delete
- k8s-ci-cluster-dcb98b0ba0ac5801fe1b47555bcddeadccc49832-1687 pulumi:pulumi:Stack delete
Looks like ProgramTest defaults to enabling t.Parallel(), and this defeats the purpose of prewarm phase in provider benchmarks. The intent of prewarm was to do a no-op to make sure dependencies are downloaded and avoid measuring that overhead as part of the benchmark. This was not the case though because prewarm was running in parallel with the actual benchmark.
After the change, each individual benchmark first does unmeasured prewarm, and then does the measurement. In the future we may want to measure prewarm as it is interesting for user experience as well.