Loren Gordon

@pliniodng That's pretty much our exact scenario also, a central repo that manages the IaC for all CodeBuild jobs, with buildspecs managed separately by each other repo. Updating the image...

@bryantbiggs One thing that occurs to me that might help with this, would be to use `depends_on` in the _output_ for the `cluster_endpoint`, forcing it to wait until the access...

@bryantbiggs Yes absolutely, using depends_on in an output is 100% valid, and used for exactly this kind of timing issue where an output for a single resource isn't _fully_ available...

They have a doc on it now also: https://developer.hashicorp.com/terraform/language/values/outputs#depends_on-explicit-output-dependencies

Fyi, I am testing a PR to see if I can cleanly apply and destroy the "complete" test on the blueprints-addons project without manually using state rm and target... So...

@bryantbiggs Done, see https://github.com/terraform-aws-modules/terraform-aws-eks/pull/3000! (And I got the 3000th pr, woot!)

Hmm, that's tricky. Is there _any_ host where you could push the RPM where it is accessible by either HTTP (not TLS), or HTTPS such that the network path bypasses...

I assume you're using [this cloud-init module](https://cloudinit.readthedocs.io/en/latest/reference/modules.html#ca-certificates) to update certs? We hadn't considered a mechanism to further customize the build quite like that, but I can understand the use case....

@evandena Would you be interested in working on a PR that implements a solution for this use case? I don't really see us getting to it any time soon... But...

go codebuild go