Run azure steps in parallel


it would really help me out a lot if I could deploy my 10 or so azure steps in parallel in some way. Right now each azure cloud service upgrade takes about 7-8 minutes, and since they run one at a time it takes a long time to run the upgrade. Most of the time is just sitting and waiting for the azure service to report back that it’s done. If there was a way I could deploy to these machines in parallel that would be amazing. Even if it meant spinning up another tentacle that could take half of the steps or something. I found this on uservoice but not much activity on it:


We have run into this as well - we have a workflow whereby we have an Octopus project for each Azure cloud service we have configured, all of them for several different environments (environment = Azure subscription here). It would be nice if, in parallel, we could theoretically run a deployment for every service into every environment. To do that currently, though, we need (# services) * (# environments) tentacles, which becomes hairy to manage very quickly.

The user voice link shows this feature as planned, is there an ETA on that? Would it be worthwhile/reasonably straightforward to make this change ourselves if we discussed a source code license with you? Or are we stuck managing heaps of tentacles when it comes to supporting parallel Azure deployments?

One potential way would be to allow the azure deployment steps to be issued async, so the step won’t sit and wait for them. Then add a new step type that will wait for all of the async steps. I figured that could be easier than built a whole parallel-step system. For my environment I wouldn’t mind multiple tentacles that much (even if I think it would be overkill), but currently it would execute the same steps on all of them so not very useful (or I would have to go and assign each step to each tentacle, messy).

Hi All,

Thanks for your feedback and comments. I have updated the UserVoice suggestion to note this thread when the design process is happening.
As you can see from our Roadmap this planned feature is still a few versions away, so we could be talking a bit of a wait, but please do update the UserVoice with relevant comments.