Capacity Tier drive appearing incorrectly as cache tier after replacement in vSAN OSA
search cancel

Capacity Tier drive appearing incorrectly as cache tier after replacement in vSAN OSA

book

Article ID: 449631

calendar_today

Updated On:

Products

VMware vSAN

Issue/Introduction

After replacing a failed capacity tier drive in a vSAN Original Storage Architecture (OSA) environment, the new drive may show as "Unclaimed" or incorrectly appear as a Cache Tier option. Standard disk group recreation may fail due to metadata locks.

  • Replaced drive is visible in "vSAN Cluster Disk" view but missing from host summary view.
  • CLI removal fails with: Unable to remove device: Failed to get VsanInfo operation lock for diskOpLock
  • Logs show: Individual capacity disk is not allowed to be decommissioned from a diskgroup-level dedup disk group

Environment

  • VMware ESXi 8.0 Update 3 (build-25429389) and higher
  • vSAN 8.0 OSA Mode
  • Deduplication and Compression enabled

Cause

In vSAN 8.0 OSA with Deduplication enabled, stale metadata in the Cluster Monitoring and Membership Directory Service (CMMDS) can prevent correct tier identification. The vsanmgmtd process may hold a persistent lock on the disk operation, preventing manual removal until the lock file is cleared and the service is cycled.

Resolution

  1. Log in to affected ESXi host via SSH as root.
  2. Identify stale UUIDs using the following command:
    vdq -i
    Note any device reporting a UUID instead of a naa.#### ID.
  3. Stop the vSAN Management service:
    /etc/init.d/vsanmgmtd stop
  4. Remove stale lock file:
    rm -f /tmp/.vsanDiskOpLock.lock.LOCK
  5. Manually delete the stale CMMDS entry for the identified UUID:
    cmmds-tool delete -u ####-####-####-####-####
  6. Restart management services:
    /etc/init.d/vsanmgmtd start
    /etc/init.d/hostd restart
  7. Clear partition labels on the new drive:
    esxcli storage core device partition label clear -d naa.####
  8. Rescan storage adapters:
    esxcli storage core adapter rescan --all
  9. Rebuild Disk Group via vSphere Client.

Additional Information

Note: Stopping hostd will temporarily disconnect the host from vCenter, though it does not impact running virtual machines.