Hello, I'm finally back to working with Pulumi :ta...
# general
n
Hello, I'm finally back to working with Pulumi šŸŽ‰ pulumipus flying pink cape after a 3-year stint with TFC (poop emoji here) It's good to be back, but I'm also a bit rusty with the CI/CD and Deployment options, I've encountered out dated docs, and I still have several challenges with onboarding my organization; I'll create a thread on this post so I wouldn't flood the channel
Questions I'm not sure about: • Difference between a stack Integration settings, and Deployment Settings (https://app.pulumi.com/org-name/project-name/stack-name/settings/integrations vs (https://app.pulumi.com/org-name/project-name/stack-name/settings/deploy) Essentially, what is the relationship between the Github pulumi app with deployments enabled and the github actions workflows, it feels like both docs and ui are sending me to do both and it feels a bit ambiguous
• The VCS Integration assistant promotes outdated github actions: ā—¦
pulumi/actions@v2
—>
pulumi/actions@v6
ā—¦
actions/checkout@v2
—>
actions/checkout@v6
ā—¦
softprops/turnstyle@v1
—>
softprops/turnstyle@v3
ā—¦
actions/setup-python@v2
—>
actions/setup-python@v6
• Missing workflow permissions - either at global or job level:
Copy code
permissions:
      actions: read # for turnstyle
      contents: read
• Out dated python3 version: ā—¦
python-version: 3.6.x
—>
python-version: 3.13.x
Pulumi docs are lacking deployment pattern examples, for monorepo multiple projects, with CI/CD for each project/stack combo
it is only mentioned
s
This is excellent feedback! Thank you for taking the time to write it out! Funny you should mention - we've been talking about that very ambiguity in the docs WRT the GH App/Action and Deployments. The GH App will get you things the
pulumi preview
output as a comment in a PR. I believe it's required in order to use Deployments as well. The GH Action is basically the Pulumi CLI with a well-defined interface so that you don't have to Bash your way through Pulumi commands. We are aware of the difficulties using Deployments at scale (because they are per-stack) and actively working on solutions for that problem.
šŸ’Ŗ 1
šŸ˜„ 1
n
Glad I could help šŸ™‚ I have a scheduled meeting today with Jack Delahunt about onboarding, we'll get something going.. I've tried enabling both GH Action and the GH App w/ deployments, and I confirmed that indeed it runs pulumi preview twice, Essentially one runs as expected in an GH action runner, and another preview runs on Pulumi cloud 'deployment agent pool' This also means that we are querying the resources cloud state in our AWS twice (for the refresh) 😵 (I'm already thinking about future API rate limits šŸ˜…)
I also found that this resource: https://www.pulumi.com/registry/packages/eks/api-docs/nodegroupv2/#inputs has a seemingly duplicate input: (python)
Copy code
encrypt_root_block_device=True
node_root_volume_encrypted=True
Docs:
Copy code
node_root_volume_encrypted: Whether to encrypt a cluster node's root volume. Defaults to false.

encrypt_root_block_device: Encrypt the root block device of the nodes in the node group.
s
My very, very general, squishy sense is that hitting API rate limits is a relatively rare thing among our customers. I have some customers who are pretty heavy duty users of some AWS APIs and they don't typically hit rate limits. If it does become an issue, you can probably work with your AWS team to get your rate limit increased. Don't remember if that's a support request or something you would do through a TAM.
šŸ‘ 1