Does anybody here use a self-managed pulumi state? I am having some difficulty switching from pulumi cloud to using an s3 bucket for the state.

aws.lambda.CallbackFunction

does not seem to work with the s3 bucket for state, I think because it cannot be serialized into json. Is this correct or do I just have it configured wrong? Changing everything into zip files for the lambda functions is difficult because, the way that the backend deployment code is written it will be very difficult to figure out which dependencies to package for functions. Especially since most lambda functions are created programatically in a big loop with many different objects and functions involved.

hey Connor, sad to hear you’re migrating off the pulumi cloud backend. That said, there shouldn’t be any difference for lambda functions. What error are you seeing?

I haven't used

aws.lambda.CallbackFunction

specifically, but we just moved from pulumi cloud to self-managed state and didn't have any issues across a dozen or so stacks and hundreds of resources.

I actually have it working fine now, I have no idea why it was performing that way before. I am noticing that deployment seems to be much slower using an s3 bucket. Are there any settings that can help with this? I’ve tried increasing maxParallel, which seems like it might help a little. I’m not sure what the bottleneck would be here.

I am noticing that deployment seems to be much slower using an s3 bucket.

How many resources do you have in your stack? Were all the resources slower to deploy?

I’ve tried increasing maxParallel, which seems like it might help a little. I’m not sure what the bottleneck would be here.

Interesting. What value did you use?

I tried 10 and also 3, neither made a significant difference after testing it again.

I believe the default value is max parallelism already.