We sometimes run into a task that takes forever (runs for days and days), usually a health check, but not always.
When attempting to cancel the task the task stays in the “Cancelling” status for days, or hours.
We have had tasks stay in the canceling state across upgrades, and server reboots in a HA cluster. Often times this creates a mini production outage because we are unable to get octopus to do certain things while it is stuck in cancel task mode. We can’t drain server nodes, or take them offline, or anything.
The expected behavior is that when you cancel a task it cancels soon after you instruct it to.