05/04/2021, 11:59 AM
Hello, we have a ci/cd pipeline (monorepo) where we build frontend, backend and some infra, each of those have their own stack. To speed things up, we build in parallel (codebuild/codepipeline) fe in one build, be in one build and infra in another build. We did notice that the infra stack (which is referenced by fe/be) sometimes is null. I assume that happens because we use s3 as backend and possibly that's locked when when doing a pulumi up in a different process. But how else could we structure this? Are we using the tooling in a wrong way?