Best Practices when Upgrading Tamr

In addition to taking backups prior to starting the process of upgrading Tamr, there are a number of other “good practices” that one should follow.

Actions to take before upgrading

1. If starting from version 2020.021.0 or later, run the CleanupIncompletelyDeletedProjects maintenance utility. Please see documentation here.
2. Verify that ulimit and vm.max_map_count are set correctly for the target version For more information on how to verify, please see the documentation here.
3. Make sure there is at least 30-40% of free disk space available on the instance to accommodate for Backups. And, Elasticsearch will not allocate shards if there is more than 85% of disk space utilized. Please go to the documentation here for the commands to check the disk space

Pitfalls to avoid when you run into issues while upgrading

1. If Tamr times out when starting up, Do NOT stop it and restart it. It might be still running upgrade scripts. Give it at least 20 minutes. You can try the health check API to see what is going on. You can also check the unify.log and see if progress is being made. Interrupting it can break the system and/or force you to rerun the upgrade.
2. If the upgrade fails with an ES problem, Do NOT immediately clear ES. Look at the ES logs and troubleshoot the underlying issue. Then rerun the upgrade with --rerun.