We’re using Octopus on-premise, v2023.2. We recently noticed in our automated we scale up of multiple EC2 instances for one or more AWS Autoscaling Groups, the Calamari update tasks that we kick off for each new deployment target do not run in parallel. Our recollection is they used to do this. But now if we scale up 4 ASGs adding 10 EC2 instances to each, when there’s plenty of tasks bandwidth, still Octopus creates a huge delay by forcing each of these Calamari update tasks to run one after the other instead of all at the same time. Since, by definition these tasks are per machine, is this a bug? Please get us back to where these tasks can run in batch all at the same time.
Hi Bob,
Thanks for the interesting query on upgrades.
I did a bit of digging on this one and while we have looked into this quite a lot, there are many conflicts in the upgrade logic which make this unfortunately a difficult feature to implement.
Our current approach is to match the number of logical processors Octopus is running with as a task cap for updates, so maybe there is some room there to play with the number of CPUs etc allocated to Octopus:
So we have had to go back to the drawing board to get a better approach and its definitely still a priority feature for us.