Single tentacle upgrade failure fails upgrade for all 40+ others (Why are tentacle upgrades transactional?)

On multiple occasions, Octopus may fail updating 1-2 tentacles out of a large group of many. The environments page then shows all of the tentacles/hosts as still needing to be updated. Is this is UI issue or have the upgrades on the successful hosts been rolled back? The UI indicates that the ‘Upgrade all machines’ button is a single transaction and if a single upgrade fails, the entire task is rolled back. This is either confusing or a bad implementation, especially given the upgrades are mandatory. Please advise. Thanks.

Hi Rick,

Thanks for getting in touch! We never roll back upgrades. It’s mostly a timeout issue and a roll on UI issue.
If a few Tentacle restart messages don’t get through it never knows they completed upgrading. We are going to make this process work better in the future.
If you run a health check it will also check the Tentacle versions and should show they all completed their upgrade correctly.

Hope that helps!