Summary
During a cluster upgrade, the list of hosts to upgrade comes from different sources. This is important to remember when looking into failed upgrades or other unexpected upgrade issues.
Solution
In addition to the host list in the Cluster Manager database, one of two files exist on the Cluster Manager machine, c:/ec_nodelist and/or c:/ec_nodelist_mega (Windows) or /var/tmp/ec_nodelist (Unix). These files are the sum of all hosts known to the cluster and act as a backup to the upgrade process.
If hosts are moved manually, this list can become out of sync with the actual cluster. Before upgrading, you can remove these files from the Cluster Manager machine and the upgrade will not be attempted on those extra machines.
This article is part of our Knowledge Base and is provided for guidance-based purposes only. The solutions or workarounds described here are not officially supported by CloudBees and may not be applicable in all environments. Use at your own discretion, and test changes in a safe environment before applying them to production systems.