Yesterday I got a new error:

error: [409] Conflict: Another update is currently in progress.

We have a small team… nobody else was running updates. 🤔 So I (boldly?) decided to run:

pulumi cancel
# then
pulumi up

but that seemed to make things worse?

warning: Attempting to deploy or update resources with 1 pending operations from previous deployment.
  * urn:pulumi:my-stack::my-app::aws:ec2/instance:Instance::my-instance, interrupted while creating


error: Detected that /path/to/user/.pulumi/plugins/resource-aws-v7.23.0/pulumi-resource-aws exited prematurely.
           This is *always* a bug in the provider. Please report the issue to the provider author as appropriate.
           To assist with debugging we have dumped the STDOUT and STDERR streams of the plugin:

It looked like it was hanging on EC2 creation. I think I had a problem with the user-data preventing proper launch 😬 So then I followed the prompting to run

pulumi refresh

To cancel pending CREATE ops… eventually

pulumi up

came up clean, but now I have 4 or 5 instances of each EC2 in the stack 😮 Was this caused by my messed up user-data? Or was this a bug in the provider (as the error mentioned)? Or something else? It seems to be all cleaned up now (?) but I’m wary I may have other duplicated resources lurking in my infra now. I don’t know if it’s relevant, but somewhere in the middle of that I updated

pulumi

(on Mac OS X using

brew

):

3.230.0 -> 3.231.0

any thoughts?

Was this caused by my messed up user-data? Or was this a bug in the provider (as the error mentioned)? Or something else? It seems to be all cleaned up now (?) but I’m wary I may have other duplicated resources lurking in my infra now.

Looks like a provider issue. Worth raising this on the aws provider repo for them to look at. Even with bad user data the provider shouldn't exit prematurely.

I guess it’s a known issue. This also explains why I ran into this problem whereas teammates did not (they happened to pull in a later/patched version of the package).

make sure deps are updated, and then generally

pulumi up

will just pick up the new versions. Is it stuck on the old version for deleting things, or other operations?

I’m not sure which operation it’s stuck on — it just says “resource-aws-v7.23.0/pulumi-resource-aws exited prematurely”. And at this point I’m not 100% it’s due to this specific version of Python `pulumi-aws`… it could very well be something whacked out on my local python install (typical python “fun stuff” where it can’t find which instance is running or its packages). I’ve been trying ops with

--parallel 1

to see if that affects anything, but so far, nothing. Another dev is working on the stack today, so I may just steer clear for a day.