Article number
000004173
Affected Versions
All
Source Hypervisor
All

Sites Show Disconnected But RPO Is Healthy And Checkpoints Accruing As If All Is Well

Viewed 264 times

Root Cause

This is a code bug with the new Alert Persistence feature that results in stale/cleared alerts remaining in the ZVM Database. After a ZVM service restart (or ZVM reboot), these "stale" alerts are reported back to the ZVM service by the ZVM Database and the GUI lists these alerts as active again as if they were active prior to restart of the service (which they were not).

Symptoms

Site disconnection alerts are present on a system running 7.0. However, replication appears to be fine as RPO is healthy and checkpoints are accruing.

There may be many other alerts as well but all checks prove that these alerts, like the site disconnection alerts mentioned above, appear to be benign.

DR Operations such as Failover or Move as well as Edit VPG may not be possible from the Protected Site.

Solution

Workaround


1) The following SQL queries should be run on each ZVM database in the pair seeing the issue:

NOTE: Kindly follow the typical best practices of backing up each ZVM DB prior to making any edits.

truncate table AlertStateInfoStorageObjectVpgs

delete AlertStateInfoStorageObject



2) Then, confirm each table is now clear via the below SQL queries:


select * from AlertStateInfoStorageObjectVpgs

select * from AlertStateInfoStorageObject



3) Monitor the UI to ensure only legitimate alerts are enabled (i.e. you should not see site disconnection alerts after this unless they are actually disconnected).


Permanent Fix

A fix was implemented into 7.0U1. Hence, upgrading to this code version will avoid this issue from occurring again.