Avi Controller Upgrade Fails with UPGRADE_FSM_ABORTED Due to Patch Version Mismatch on One Controller Node
book
Article ID: 422851
calendar_today
Updated On:
Products
VMware Avi Load Balancer
Issue/Introduction
The Avi Controller upgrade fails after the controller reboot during a cluster upgrade.
The following behavior is observed:
Upgrade status shows UPGRADE_FSM_ABORTED
Error indicates ["Cluster nodes need to be homogeneous regarding memory, CPU, disk, version, patch version, and FIPS mode."]
The patch is applied on the leader and follower nodes, but one follower node reports the patch version as Empty, resulting in a patch version mismatch across the controller cluster.
Errors are logged in /var/lib/avi/log/upgrade-coordinator.log on the Controller Leader node (check both current and previous partitions).
Due to this mismatch in patch status across controller nodes, the upgrade does not complete.
Environment
Avi Load Balancer
Cause
Patch image transfer from the leader to one follower controller node fails due to a connectivity interruption.
The patch package on the affected follower node remains incomplete or corrupted.
The connection lost between controller nodes during patch image transfer is logged in /var/lib/avi/log/remote_task_manager.INFO on the Controller Leader node.
Resolution
After the controller upgrade fails, identify the controller node where the patch version shows as empty by reviewing /var/lib/avi/log/upgrade-coordinator.log
Log in to the affected controller node and verify the patch status in the old partition; if the patch version shows as null, the patch is not applied.
Validate the MD5 checksum of the controller patch package located at /host/pkgs/<patch_directory>/ on the follower controller node.
Run the same MD5 command on the leader controller node and compare the values; if the checksums do not match, the patch package on the follower node is corrupted.
Remove the existing patch package from the Avi Controller GUI on the leader node, upload a fresh patch package, and restart the controller upgrade process to complete it successfully.