Unable to delete inaccessible vSAN object
search cancel

Unable to delete inaccessible vSAN object

book

Article ID: 385260

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

Symptoms:

  • Attempting to delete the inaccessible vSAN object from vSAN Skyline Health > Data > Object Health > "Purge Inaccessible VM Swap Objects" (for vm swap objects) or with objtool command completes with no errors, but the inaccessible vSAN object does not get deleted.

  • Owner abdication of the object does not change the owner of the object.
    Note: There can be times when the owner does change, but after attempting to delete the object the owner changes back to the original one.

Environment

VMware vSAN 8.x (ESA)

Cause

  • The inaccessible vSAN object is unable to be deleted as the components which make it up has reached their max "num prepares" value of 224.

  • If this occurs no further operations for the object will proceed, such as deletion.

  • /var/run/log/epd.log on ESXi host reports the below for the inaccessible object:

    YYYY-MM-DDTHH:MM:SS No(13) epd[2100587]: EPDGetEntryStalenessForSimplifiedDC: Entry ########-####-####-####-############ is stale
    YYYY-MM-DDTHH:MM:SS No(13) epd[2100587]: EPDGetEntryStalenessForSimplifiedDC: Entry ########-####-####-####-############ is stale

  •  vsanTraceReader.log reports the below for the inaccessible objects:

    2025-05-05T15:48:23.046675 [202315684] [cpu51] [3c83f1c5 COMP deleteComponent] DOMTraceOperationNotifyResponse:3830: {'op': 0x45bd8a7dd740, 'obj': 0x45dcbc5c58c0, 'objUuid': <disk_uuid>, 'opState': 'DOM_OP_STATE_COMPLETE_TASK', 'opActivator': 'LSOM', 'opTaskIdx': 0, 'resultCode': 'VMK_LIMIT_EXCEEDED'}

  • The value for "Num prepares" for an absent component is 224:

    1. Choose an absent component for the inaccessible vSAN object, make a note of it's UUID.

    2. For this component, make a note of the "Disk UUID" & "Host name". To get components of the inaccessible object, run the command: esxcli vsan debug object list -u <inaccessible_object_uuid>

      Example of a component from an inaccessible object hit with this issue:
      Component: <component_uuid>
             Component State: ABSENT,  Address Space(B): ####### (####GB),  Disk UUID: <disk_uuid>,  Disk Name: <disk_name>
             Votes: 2,  Capacity Used(B): 0 (0.00GB),  Physical Capacity Used(B): ##### (###GB),  Total 4K Blocks Used(B): 0 (0.00GB),  Host Name: <hostname>

    3. Log into SSH of this host and run this command to verify the num prepares value: vsish -e get /vmkModules/lsom/lsom2/disks/<disk_uuid>/components/<component_uuid>/info
      Output similar to:
      {

         Partition UUID:<component_uuid>
         Partition type:capacity
         Component pointer:0xxxxxxxxxxxxx
         Component ID:########
         Comp table page Idx:1
         Comp table entry Idx:2
         State:4
         Num prepares:224          <<<<<<<<<<<<<<
         .
         .

Resolution

The fix for this issue is available in ESXi 8.0U2b (24280767).

 

Note: This situation can also arise for objects which were inaccessible when the ESXi host was on a version below 8.0U2b and the hosts are now on version 8.0U2b or above. A workaround is available for this scenario. Please contact Broadcom Support for further investigation and assistance.

Additional Information

Powered by Wolken Software