Some of the VMs disks fail to load -Error Unable to enumerate all disks

Products

• VMware vSphere ESXi

Issue/Introduction

The following is observed in the environment:

• When trying to power on a VM error "Unable to enumerate all disks" is received
  
• Not all vmdks show the correct size for the VM
  
• The vmdks showing no size are RDMs
• Pure storage replication is in use
• All paths to the RDMs via Host > Configure > Storage Adapter > Adapter > Paths show as Dead
• vmkernel.log following messages are seen:
  
  

  [YYYY-MM-DDTHH:MM:SS] Wa(180) vmkwarning: cpu52:2098139)WARNING: NMP: nmp_CallRegisterDevice:1293: Device, seen through path vmhba#:C#:T#:L## is not registered (no active paths)

  [YYYY-MM-DDTHH:MM:SS] In(182) vmkernel: cpu53:2098663)NMP: nmp_ThrottleLogForDevice:3893: Cmd 0xa3 (0x45d9c1b1e400, 0) to dev "Unregistered Device" on path "vmhba#:C#:T#:L##" Failed:

  [YYYY-MM-DDTHH:MM:SS] In(182) vmkernel: cpu53:2098663)NMP: nmp_ThrottleLogForDevice:3898: H:0x0 D:0x2 P:0x0 Valid sense data: 0x2 0x3e 0x1. Act:NONE. cmdId.initiator=0x45399aa1bc48 CmdSN 0x0

  [YYYY-MM-DDTHH:MM:SS] Wa(180) vmkwarning: cpu52:2098139)WARNING: VMW_SATP_ALUA: satp_alua_issueCommandOnPath:1068: Path "vmhba#:C#:T#:L##" determined to be in unexpected NOT READY state when probed (0x2/0x3e/0x1).

  [YYYY-MM-DDTHH:MM:SS] Wa(180) vmkwarning: cpu52:2098139)WARNING: VMW_SATP_ALUA: satp_alua_issueCommandOnPath:1068: Path "vmhba#:C#:T#:L##" determined to be in unexpected NOT READY state when probed (0x2/0x3e/0x1).

0x2 - NOT READY
0x3e/01 - LOGICAL UNIT FAILURE

 

Environment

vSphere ESXi (All Versions)

Pure Storage Array

Cause

This is due to the RDM LUNs are configured for replication via the Pure Storage Array. The target LUNs, in this case the vmdks showing 0 in size, are demoted to read only so the LUNs are unable to accept new writes. While these LUNs are configured for replication and demoted to read only the paths show as dead via the storage adapter until they're promoted back to a read/write state.

As per Pure document Preparing the VMware Environment for a Pod Demotion the RDMs should have been removed from the VM while in this state.

Resolution

To resolve this do the following:

Remove the RDM vmdks from the VM

Optional: To confirm there are no issues with the RDM LUNs (Steps to be performed by storage team/Pure engineer)

1. Split the storage
2. Promote the RDM LUNs to read/write
3. Rescan storage for the cluster to reinitialize the paths
4. The vmdks should now show their respective proper size