Known Issues

v1.0.1

The restore might place the restored PVs in a vSphere datastore which is imcompatible with the StorageClass specified in the restored PVC, as we don't rely on the StorageClass for the PV placement on restore in the release v1.0.0. A fix to the PV placement based on the StorageClass can be expected in a future release.

Restore fails if the K8s node VMs are placed in any sub-folders of vSphere VM inventory.
1. Workaround: move node VMs to the top level of VM inventory. (for users of release v1.0.0)
2. Solution: the issue was resolved in the master branch at commit 1c3dd7e20c3198e4a94a5a17874d4c474b48c2e2.
The restore might place the restored PVs in a vSphere datastore which is imcompatible with the StorageClass specified in the restored PVC, as we don't rely on the StorageClass for the PV placement on restore in the release v1.0.0. A fix to the PV placement based on the StorageClass can be expected in a future release.
The plugin v1.0.0 has issues with some variants of Kubernetes on vSphere.
1. In a TKG cluster, the vSphere CSI credential secret, i.e., VC credential, is named csi-vsphere-config in the kube-system namespace, which is different than the expected vsphere-config-secret in the kube-system namespace. As a workaround, add another secret, vsphere-config-secret, with VC credential available in the secret csi-vsphere-config. Some more tips on using this workaround: Issue #63.
  1. Make sure to use quotes to wrap around the value of each key-value pair in the csi-vsphere.conf when constructing vsphere-config-secret.
  2. Make sure to add one more key-value pair, port = "443", under the VirtualCenter section. As far as we know, the port key is absent from the csi-vsphere.conf of the secret csi-vsphere-config in TKG.