VMware Aria Operations for Logs agent auto-update behavior when configured with multiple destinations

Products

• VCF Operations

Issue/Introduction

When an agent is configured to forward logs to multiple Aria Operations for Logs destinations that are running different versions, the agent may fail to auto-update or appear "stuck" on a lower version.

Environment

VMware Aria Operations for Logs 8.x

Cause

This is an expected agent's behaviour.

• The agent creates a single, shared UpdateHandler for all configured destinations.
• While each destination independently polls its server, they all write to the same shared version/download state.
• The logic (in UpdateHelper.cpp) was originally designed for load-balanced nodes within a single cluster where all nodes share the same version. It was not intended for destinations at different version levels.
• Race condition: The agent determines which version to use based on whichever destination sends its status response first. Because the agent manages updates through a single process for all destinations, it locks into the version offered by that first responder and ignores other updates until the cycle completes. If a destination with an older version happens to respond first, the agent often gets "pinned" to that older version instead of upgrading to a newer one.

Resolution

To ensure predictable auto-update behavior in multi-destination environments:

1. Version Alignment: Ensure all destination Aria Operations for Logs clusters/instances are running the same version.
2. Manual Management: If destinations must remain on different versions:
   • Disable the auto-update feature for that specific agent in the liagent.ini file:
      

     [update]auto_update=no

   • Manually upgrade the agent to the desired version using the MSI (Windows) or RPM/DEB/BIN (Linux) package.