we are running Octopus Deploy in HA mode (2 nodes currently) using the official Docker image. It is hosted on an AWS ECS cluster. Is it safe to upgrade to a newer version by just updating the ECS task definition to use a new image, i.e. keeping multiple nodes during the upgrade process? My only concern is what if both nodes try to apply update to the database?
Thanks for getting in touch!
When running HA and performing an upgrade it is best to take the additional nodes offline and upgrade a single node that will then perform the database upgrades. Once that is done any additional nodes can be upgraded together.
Thanks for your quick response. The problem is the whole deployment is completely automated. In ideal case I could just bump the version number and trigger a redeploy. It works, I tested it this way multiple times and never had issues. But I still felt like I should confirm if there’s any risk this could lead to database corruption.
It sounds like except the version number bump I’ll also need to modify the desired count of instances in the ECS service to 1, run a redeploy, wait for the upgrade, change the number of instances back, run another redeploy. A bit more annoying so wanted to avoid this process unless necessary, but sounds like I’ll need to go that path.
Thank you again.
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.