No matter how you like to participate in developer communities, Pulumi wants to meet you there. If you want to meet other Pulumi users to share use-cases and best practices, contribute code or documentation, see us at an event, or just tell a story about something cool you did with Pulumi, you are part of our community.

Pulumi Community

I think it might be related to us not waiting for a promise to finish or something somewhere in the code. But that shouldn't impact two different runs as I see it. Ideally the pulumi client should be "stateless", so it failing once shouldn't impact the next call which seems to be happening now.

&gt; Ideally the pulumi client should be "stateless",
Ah it's not. If your using inline stacks there is global state inside the pulumi node sdk so you can't run operations in parallel.

ok, so that explains the first part of why I can experience what I see, that is, if it fails one time the next can also fail... but it doesn't explain why I do get the error even though everything has succeeded.

And I thought/question. Does that state need to be in the SDK? Shouldn't in that case be a call against the API to verify if something is running or not? I mean, if you create an web api using automation there might other nodes trying to run operations at the same time.

Not having the state as a global in the SDK would require a large api change to pass the deployment context through everywhere (because node doesn't have async locals). The go SDK is written like this, maybe 4.0 we could revisit it.

That's fine, but I still don't understand why I get an "One or more errors occurred" even though all the resources have been provisioned, and that it only happens after on deploy has failed first.

My guess is that the async tracking the engine does is picking up some left over asyncs from the first update and thinking their part of the other updates. Do you get any more logs than just "One or more errors occurred"?

That is the only log I get, so it is quite hard to debug. I do think I am awaiting all async, but must be missing something. 

Just to check this is with the nodejs automation api and using inline programs?

yes, it is with an inline program using the nodejs api

I can check if I can make a smaller repro as well. 