When polling tentacle goes offline during deployment, the current step runs indefinitely

When polling tentacle goes offline during deploy, it runs indefinitely.

Octopus Deploy (tentacle and server)

Given a tentacle goes offline and after starting a deployment the active deployment step will remain running indefinitely. Maybe there should be an option to trigger a failure after a timeout is reached, configured per step. This could also be automatic based on the average execution times that are already measured.

In my particular use case I’m deploying software to a group of 5 test machines configured with polling tentacles, and one lost connectivity for extended periods (remote site) a few times. In all cases, 4 machines completed successfully within the expected 5 mins, but the 5th machine that went offline the current step runs indefinitely, so the job never completes or fails. The longest was 21 hours mid post-deploy. Per tentacle health checks, it was indeed offline for 21 hours.

My concern is that in production I’ll be deploying to much larger environments (50+ machines per unique environment), and there is a probability that certain machines may go offline, so the ability to force a failure in order to take advantage of the guided failure features is very important.

Hi Arri,

Interesting points; currently the timeout is set at 2 days, because of the need to support e.g. deployment to occasionally-connected corporate laptops.

Configuring this limit as a user sounds like a worthwhile enhancement, I’ve added an issue to track it here: https://github.com/OctopusDeploy/Issues/issues/526