Network issue during velero backup

Products

• VMware Tanzu Kubernetes Grid Integrated Edition

Issue/Introduction

Worker node experiences network connectivity issues for 30-40 seconds while velero is performing a backup that includes taking a CSI snapshot of a PV.

The network connectivity issues observed could include "no route to host" and "context deadline exceeded" from pods and process running on the Worker node.

The VM kernel logs show that the network interface is hanging

vmxnet3 0000:03:00.0 eth0: tx hang
vmxnet3 0000:03:00.0 eth0: resetting

Environment

Any Kubernetes cluster running velero backup with CSI snapshots.

ESX 7.0.3

Resolution

Known issue in ESX and fix will be available in future releases as below:

• 7.0.3 P10
• 8.0.3.0 P06
• 9.0.0.0

Additional Information

KB for ESX Build numbers and versions