Problems with setting up maximum number of concurrent steps execution




I am trying to change number of executed steps, now the limit as I understand is 10 and I want to change it to 6

I was trying to use variable Octopus.Action.MaxParallelism and set it on project level but without any success.

already checked link

(Daniel Fischer) #3


Thanks for getting in touch! I may need to get some more information about your issue. I think the best course of action is to have you attach the raw deployment log file with variable logging enabled, for the task where you are having this problem.

The log file will give me a contextual idea of what is happening during your deployment and what could be going wrong.

My current assumption is that your deployment is still running at 10 tasks in parallel instead of 6. Is this correct?

Are you able to confirm that you created a new release after adding this variable? Any variables are associated with a release, so new variables will need a new release to be picked up in your deployment.

Looking forward to hearing from you. :slight_smile:

Best regards,


Yes deployment is still running 10 tasks in parallel instead of 6

I have attached logs of simple release execution of test project where behaviour is exact that same as in one which is used for our tool
ServerTasks-7545.log.txt (44.2 KB)

(Daniel Fischer) #5


Thanks for getting back and confirming that. Unfortunately the log file does not have variable logging enabled which is required to troubleshoot this issue.

Would you be able to add the following variables to your test project and create another release + Deployment, then attach this new log?

Variable Value
OctopusPrintVariables True
OctopusPrintEvaluatedVariables True

These project variables will output variable information from the deployment which is usually hidden.

Looking forward to hearing from you.

Best regards,


Hi, new logs with enabled variablesServerTasks-7938.log.txt (897.6 KB)

(Daniel Fischer) #7


Thanks for getting back with the variable logs. Though it think I know whats going on here. I may have initially misinterpreted your issue, the Octopus.Aciton.MaxParallelism variable is responsible for how many machines/targets each step will run on in parallel. It looks like you are trying to specify how many deployment steps should be executed by octopus in parallel.

To configure this, you will need to put your steps into child steps. This have Octopus run the child steps in batches.

The documentation I linked to has a lot more information on this.

Let me know if you have any thoughts here, or if I have misinterpreted your issue still. :slight_smile:

Best regards,


That’s a pity, I was hoping to use MaxParallelism to make my deployments more elastic.
Now we are running multiple deployment steps in batches 6 in parallel then next 6 are waiting for first one set to finish and that same for next series. Now we have 36 deployment steps and those steps deployment time is different at each execution(it is related to the jobe executed by particular msvc, sometimes we have to wait before shutting it down). So sometimes entire series will update smoothly and sometimes we have to wait for single msvc to finish his upgrade before starting next one, and that causes more or less 2h upgrade of single zone.
I was hoping that there could be some kind of queue from which octopus will take only 6 steps and when any of that steps will be finish then he will grab next one. I don’t think that I will be able to achieve that with child steps.

(Daniel Fischer) #9


I’m really sorry for the very late reply here. Your reply came to me just as I was taking some time off and it seems to have fallen through the cracks!

I’m just touching base to see if this is still causing you some pain.

I think you are correct in that child steps will not help in your specific scenario. However, I’m not sure Octopus currently has the ability to do exactly what you are after here.

Is your primary issue here performance or a bottleneck in your deployment? If it is, I could get some help from our performance guru’s which may be able to provide some other ideas or thoughts on how to get around this issue.

Again, I’m very sorry for this late reply. Please let me know if I am still able to help.

Best regards,

(system) #10

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.