Network Management

Multicluster HA Pool breaks the Active Agent config file in SolarWinds Platform 2025.4

When a Multicluster HA Pool is enabled and later disabled, Active Agents may retain outdated configurations, causing them to point to incorrect pollers.

First published date

11/20/2025 4:18 PM

Last published date

11/20/2025 8:15 PM

Overview

In a Multicluster HA Pool, Active Agents can randomly update their configuration files to use a different poller, even after the pool is disabled.

  • In the Web console and database, an Agent may appear assigned to poller engine X, but RDP/SSH into the Active Agent may reveal its config file is using APE Y.

  • When the pool is enabled, Active Agents correctly receive information for all APEs. However, when the pool is disabled, they are not updated to reflect that the APE is standalone.

  • Even manually reassigning the Agent to the correct APE does not guarantee the config file will remain correct.

According to the Confluence documment, in a clustered HA setup, Active Agents cannot assume they are receiving correct polling information about their own engine or backup because multiple engines and backups may be active in the pool.

Product section

Orion Platform

Cause

The Active Management Service (AMS) continues using HA pool data even after the pool is disabled, causing Active Agents to receive incorrect polling engine information and resulting in gaps in monitoring charts. 

Resolution

Active Agents are not supported under a Multicluster HA pool.

Workarounds:

Option 1: Convert to Standard HA Pool

    • Delete the Multicluster HA Pool.

    • Create a standard HA pool instead.

Option 2: Change Agents to Passive Mode

    • Modify the affected Agents to run in passive mode to prevent them from relying on incorrect polling engine data.