This question is for anyone using Pulumi in an Azure Devops pipeline with python. What are the ways you are dealing with installing pulumi_azure_native into the venv for each individual program in your pipeline? In our experience it is taking just over 2 minutes. So for example if you have 3 separate programs in a mono-repo that create a resource group, then create the necessary network pieces, then create an AKS cluster, the pipeline is spending over 6 minutes just installing dependencies not even counting the time it takes Pulumi to actually do anything.

I asked around internally, and one of our engineers had a good thought. Are you running each venv install and pulumi up in separate jobs? If you are, you might be losing the local download cache between each job, which can make it take that long. A quick SO post that might help if that's the case: https://stackoverflow.com/questions/62420695/how-to-cache-pip-packages-within-azure-pipelines

Thanks Laura. Yes we are running them as separate jobs. We are now working to convert to a combination of Docker + cache@2 to attempt to speed things up.