All paths down events during Rubrik storage array integrated backup

Products

• VMware vCenter Server

Issue/Introduction

Backups are implemented using Rubrik with Storage array integration. The backups are running successfully. However, when snapshot-based LUNs are removed from the environment, vCenter generates "All Paths Down" and "Lost Connectivity" events related to the removed snapshot LUNs. This behavior results in repeated alerts and notifications during snapshot LUN removal.

 

Cause

It is expected behavior for an ESXi host to report APD (All Paths Down) or Lost Connectivity if a LUN it still believes it has access to suddenly becomes unavailable or is removed from the storage array before the ESXi host has been gracefully informed and updated its storage inventory.

Resolution

• • The issue arises from an incorrect sequence of operations. For a smooth removal without APD events, the steps must be:

    1. Unmount the snapshot datastore from all ESXi hosts.
    2. Detach the LUN from all ESXi hosts.
    3. Perform a storage rescan on all ESXi hosts to ensure they recognize the device is no longer present.
    4. Only then, unpresent/destroy the LUN from the storage array (Pure Storage)
    
    If step 4 (removing the LUN from the array) occurs before steps 2 and 3 (detaching and rescanning on ESXi), the ESXi hosts will continue to hold references to the LUN and will generate APD messages when their attempts to access it fail.
    
    
  • DO NOT disable these alerts in vCenter/ESXi itself. These alerts are critical for detecting genuine storage outages. Engage with Rubrik and Storage array vendor to ensure their integration follows the correct, graceful detachment sequence: unmount, detach, rescan, then unpresent the LUN. This will prevent the generation of these false positive APD events at the source.