fix e2e test failure

[andyzhangx]

What happened:
It will create a volume under andy-ci resource group anyway, I checked, in the start up of this driver, it will load AZURE_CREDENTIAL_FILE file, so finally all provisioning volume job are depending on the driver itself. Is this the side effect of excluing some e2e tests? @ZeroMagic
The correct behavior should be always leveraging the deployed CSI driver other than current driver binary.

^[[32m• [SLOW TEST:118.378 seconds]^[[0m
Dynamic Provisioning
^[[90m/root/go/src/github.com/kubernetes-sigs/azuredisk-csi-driver/test/e2e/dynamic_provisioning.go:32^[[0m
  should create a volume on demand
  ^[[90m/root/go/src/github.com/kubernetes-sigs/azuredisk-csi-driver/test/e2e/dynamic_provisioning.go:55^[[0m
^[[90m------------------------------^[[0m
^[[0m[azuredisk-csi-e2e] [single-az] Pre-Provisioned^[[0m
  ^[[1m[env] should use a pre-provisioned volume and retain PV with reclaimPolicy "Retain"^[[0m
  ^[[37m/root/go/src/github.com/kubernetes-sigs/azuredisk-csi-driver/test/e2e/pre_provisioning.go:112^[[0m
[BeforeEach] [azuredisk-csi-e2e] [single-az] Pre-Provisioned
  /root/go/src/github.com/kubernetes-sigs/azuredisk-csi-driver/vendor/k8s.io/kubernetes/test/e2e/framework/framework.go:150
^[[1mSTEP^[[0m: Creating a kubernetes client
Aug  5 13:02:17.645: INFO: >>> kubeConfig: /root/.kube/config
^[[1mSTEP^[[0m: Building a namespace api object, basename azuredisk
^[[1mSTEP^[[0m: Waiting for a default service account to be provisioned in namespace
[BeforeEach] [azuredisk-csi-e2e] [single-az] Pre-Provisioned
  /root/go/src/github.com/kubernetes-sigs/azuredisk-csi-driver/test/e2e/pre_provisioning.go:60
[It] [env] should use a pre-provisioned volume and retain PV with reclaimPolicy "Retain"
  /root/go/src/github.com/kubernetes-sigs/azuredisk-csi-driver/test/e2e/pre_provisioning.go:112
^[[1mSTEP^[[0m: Successfully provisioned AzureDisk volume: "/subscriptions/.../resourceGroups/andy-ci/providers/Microsoft.Compute/disks/pre-provisioned-retain-reclaimPolicy"

^[[1mSTEP^[[0m: setting up the PV
^[[1mSTEP^[[0m: creating a PV
^[[1mSTEP^[[0m: setting up the PVC
^[[1mSTEP^[[0m: creating a PVC
^[[1mSTEP^[[0m: waiting for PVC to be in phase "Bound"
Aug  5 13:02:28.936: INFO: Waiting up to 5m0s for PersistentVolumeClaims [pvc-n99dx] to have phase Bound
Aug  5 13:02:28.952: INFO: PersistentVolumeClaim pvc-n99dx found but phase is Pending instead of Bound.
Aug  5 13:02:30.969: INFO: PersistentVolumeClaim pvc-n99dx found and phase=Bound (2.032628067s)

What you expected to happen:

How to reproduce it:

Anything else we need to know?:

Environment:

• CSI Driver version:
• Kubernetes version (use kubectl version):
• OS (e.g. from /etc/os-release):
• Kernel (e.g. uname -a):
• Install tools:
• Others:

/assign @ZeroMagic

[andyzhangx]

Below two failures are due to azure.json not consistent with kubeconfig:

Summarizing 2 Failures:

[Fail] Dynamic Provisioning [It] [env] should retain PV with reclaimPolicy "Retain"
/root/go/src/github.com/kubernetes-sigs/azuredisk-csi-driver/test/e2e/testsuites/testsuites.go:357

[Fail] [azuredisk-csi-e2e] [single-az] Pre-Provisioned [It] [env] should use a pre-provisioned volume and mount it as readOnly in a pod
/root/go/src/github.com/kubernetes-sigs/azuredisk-csi-driver/test/e2e/testsuites/testsuites.go:547

Ran 8 of 8 Specs in 1696.749 seconds
FAIL! -- 6 Passed | 2 Failed | 0 Pending | 0 Skipped
--- FAIL: TestE2E (1696.75s)
FAIL

[ZeroMagic]

Currently, we could not use just only one file to work on an AKS cluster. So We will still use azure.json in e2e test

/close

[k8s-ci-robot]

@ZeroMagic: Closing this issue.

In response to this:

Currently, we could not use just only one file to work on an AKS cluster. So We will still use azure.json in e2e test

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.