ESXi hosts disconnect from VASA Provider and lose access to vVOL datastore

Products

• VMware vSphere ESXi

Issue/Introduction

• Virtual Volume (vVOL) VASA Provider is connected in vCenter
  
  
• All ESXi hosts show as disconnected from the VASA Provider
  
  
• ESXi VM operations such as power on, snasphot and migration fail. 
  
  
• ESXi connectiivty to the VP is restored after ESXi host certificate refresh 

Environment

VMware vSphere ESXi 8.x

Cause

An issue on the VASA Provider triggered invalidation of its current session, without this being communicated to ESXi. ESXi continues to attempt to communicate via the existing session, but these attempts are rejected by the VP. 

ESXi /var/run/log/vvold.log may temporarily report "No VP available for array": 

VVold[10855649] [Originator@6876 sub=Default IpcCtx=2099842-16] VasaOp::DoRetry [#1635605]: No VP available for array (###.#######.####:#########-####-####-####-################)


Folllowed by repeated command failures with INVALID_SESSION warning:

VVold[10855676] [Originator@6876 sub=Default OpId=Session] VasaOp[#1635652] ===> FINAL FAILURE getEvents, error (INVALID_SESSION / Bad session state (Uninitialized)) VP (<VPName>) Container (<ContainerName>) timeElapsed=0 msecs (#outstanding 0).

Resolution

This issue can be resolved by re-initializing communication between the ESXi hosts and the array.

1. Via an SSH session to the ESXi host:
   /etc/init.d/vvold ssl_reset

If the issue is not resolved, additionally restart the vvold service:
/etc/init.d/vvold restart

     2. Engage VASA Provider Support to investigate what occurred on the VASA Provider to trigger invalidation of the VP session.