Ahh - yes - I was referring to debugging the deployment code (though honestly it hasn't in general been that common to need to debug those programs as when there are issues the failure modes tend to make clear whats going on). For debugging runtime callback code - that is indeed a more interesting and common case. Today, the answer is generally to use existing tools for diagnosing issues in serverless function or containers - logging, IOPipe, AWS XRay, etc. Though unfortunately, the state of the art in this space isn't particularly great. We do offer a simple

pulumi logs

command that provides an easy way to see logs from all the sources of compute in your application.