Julianne DeMars

Hey there! Just letting you know that I have just fixed and re-enabled the sanity tests for PD in go/pdcsi-oss-driver/issues/990 so if you experience failures in your sanity test PR...

I noticed there are a few more failures now so I am going to pick this up and try to fix them. ``` [Fail] Controller Service [Controller Server] [AfterEach] ControllerPublishVolume...

Okay so #1083 fixes the latter 3 of those failures, but the first one (which this bug was originally about) still exists. Note that (as @saikat-royc pointed out to me)...

Oh alrighty I see https://github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/issues/991. Do we want to abandon #1083 and work on that instead? Or go ahead with it, fix the other issue (that this bug is about)...

I summarized the failing test in question above (see steps 1-> 4), but an explanation of all the tests is at https://github.com/kubernetes-csi/csi-test/blob/master/pkg/sanity/README.md. They are supplied by CSI itself (not in...

Ah sorry - ok [here](https://github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/blob/e95d877c30895c3069b6c83bc8de30abc81b2784/test/sanity/sanity_test.go#L37) is where the tests are configured. Looks like we mock out: - [GCE](https://github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/blob/e95d877c30895c3069b6c83bc8de30abc81b2784/test/sanity/sanity_test.go#L57) - [Mounter](https://github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/blob/e95d877c30895c3069b6c83bc8de30abc81b2784/test/sanity/sanity_test.go#L62-L63) - [DeviceUtils](https://github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/blob/e95d877c30895c3069b6c83bc8de30abc81b2784/test/sanity/sanity_test.go#L62-L63) - [metadataservice](https://github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/blob/e95d877c30895c3069b6c83bc8de30abc81b2784/test/sanity/sanity_test.go#L68) - [Statter](https://github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/blob/e95d877c30895c3069b6c83bc8de30abc81b2784/test/sanity/sanity_test.go#L68) But [these](https://github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/blob/e95d877c30895c3069b6c83bc8de30abc81b2784/test/sanity/sanity_test.go#L66-L69) are real...

I'm still not sure that would help this specific test failure since the backoff occurs (as expected) within the test, which affects the cleanup in a later part of the...

They all run in the same It() for this case - [this one](https://github.com/kubernetes-csi/csi-test/blob/e3d351ac3db771d36893f3ee74df1c687a4071ca/pkg/sanity/controller.go#L1001-L1052). The problem of separate test cases affecting eachother's backoff scenarios no longers presents after your change in...

Re: [my comment above](https://github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/issues/990#issuecomment-1319636504), the problem is happening within one test, one It(). For the one failing test, within one It() we do: > 1. We create a volume ->...

Another possible solution here maybe is to make the backoff conditions RPC-specific? So that failing a publish request would not stop an unpublish request from failing, which seems to be...