Jing Xu

@SandeepPissay is there some public doc related to svmotion API for volumes in vSphere 7.0 Update 3? Thanks!

I also observe the similar issue. Some details from the log 1. volume is attached to node1 successfully 2. volume is detached to node1 successfully (but strange thing is detach...

cc @xing-yang @divyenpatel

vSphere version 7.0u3 CSI driver: 2.4.0 The test (anthos storage qualification test) is designed to move the pod to a different node and then access the same volume.

From our testing, looks 6.7u3 is ok, but 7.0 failed very often

Also another issue related to this test is since attach volume failed with "not found" error, attach_detach_controller will mark volume as uncertain, and try to detach volume when pod is...

I don't think there is datastore evacuation happening. This specific test flaky is "an existing volume should be accessible on a new node after cluster scale up" from Anthos qualification...

vsphere-csi-controller log: https://gist.github.com/jingxu97/842b05c0452d8696d98a96e56793a048 csi-attacher log: https://gist.github.com/jingxu97/0289b2277251b31a01f75b057fd47615

It turned out our system backend (maintenance activity) did issue deleting volumes in certain cases. So it is not an issue from vcenter side.

@MarkMorsing, could you provide some details on how enabling DRS on a kubernetes datastore breaks the FCD path? Thanks!