Rolling Deploys and Auto-Deploy 30 second delay

(Andrew) #1

Due to the lack of rolling deploys with the DeployRelease feature, we’ve had to create an environment per deployment target to ensure all of our projects deploy fully to 1 machine before moving to the next.
We have an application level project (i.e. MyApplication) that coordinates and deploys many other projects (i.e. MySubApp1, MySubApp2, etc).

Unfortunately, this is making deploy times unnecessarily long since we have to wait up to 30 seconds for the next Auto Deploy trigger to cause the next environment/server deploy to kick off.
(Some further background is available here:

and here

We’re currently in a rough spot because we cannot roll deploys using DeployRelease, and we cannot continue migrating more projects (which use more servers) to Octopus without dramatically impacting deployment times due to the 30 second Auto-Deploy schedule. For example, with 10 servers it can add up to 4.5 minutes just waiting for the Auto-Deploy to kick off, and 100 servers could add up to almost 50 minutes of wasted time :frowning: .

I’m hoping that one of the following is possible to allow us to continuing growing in our use of Octopus:

  1. The DeployRelease rolling deployment feature gets completed
  2. The Auto-Deploy schedule is handled via event triggers or the schedule delay can be configured
  3. Perhaps there’s another workaround for the meantime. I’ve considered looking into using the API to see how to re-invent the Auto-Deploy mechanism that immediately triggers the next deploy upon completion of another, but I’d like that to be a completely last resort.

Please let me know if there’s any progress being made on #1, if #2 is a possibility, or any suggestions for how to workaround both of them using #3 above.

Thank you!