I seem to have come across a bug? Unexpected behavior at least: I’m building a lambda function from a local zip file. I’m looping over 2 regions. But only the first region gets the updated code.

pulumi preview

etc. show that there are no changes, but when I look in the AWS dashboard I can see that one region still has the old code. This is an app with one stack that deploys to 2 region. I think the culprit is somehow this:

code=pulumi.FileArchive("path/to/lambda.zip"),

I even tried making 2 copies of the file (one for each region), but pulumi still isn’t seeing that one of the lambdas needs to be updated. 🤔 Any ideas?

Can you add a text snippet with the relevant loopy code? Is it using AWS v7 or v6?

it’s sorta this:

import pulumi
import pulumi_aws as aws
import shutil
from providers import providers
from regional.iam_role_lambda import iam_roles
from regional.s3_artifacts import s3_artifacts

cfg = pulumi.Config()
prefix = cfg.require("resourcePrefix")
regions = cfg.require_object("regions") # e.g. us-east-1, us-west-2
tags = cfg.get_object("tags") or {}

my_lambda = {}

for r in regions:
    provider = providers[r]

    # Region-specific copy of the zip;
    regional_zip_path = f"path/to/lambda_src-{r}.zip"
    shutil.copy2("path/to/lambda_src.zip", regional_zip_path)

    # The role this Lambda assumes
    role = iam_roles[r]

    # Build the Lambda function
    fn = aws.lambda_.Function(
        f"{prefix}source-builder-{r}",
        name=f"{prefix}source-builder-{r}",
        role=role.arn,
        runtime="python3.12",
        handler="index.lambda_handler",
        timeout=300,
        memory_size=256,
        code=pulumi.FileArchive(regional_zip_path),
        environment=aws.lambda_.FunctionEnvironmentArgs(
            variables={
                # ... 
            }
        ),
        tags=tags,
        opts=pulumi.ResourceOptions(
            provider=provider,
            depends_on=[
                # ... 
            ],
        ),
    )

    my_lambda[r] = fn

pulumi.export(
    "my_lambda",
    {
        r: {
            "name": my_lambda[r].name,
            "arn": my_lambda[r].arn,
        }
        for r in regions
    },
)

Ah, so already multiple FileArchive resources... hmm.

`pulumi preview” shows that there are no changes. I don’t know how long this has been happening… clearly the code got into the lambda somehow. AFAIK the providers have been working — buckets, ssm params, ddb tables etc. I only have one lambda function to deploy, so this is the only time i’ve used the FileArchive

Wondering if maybe r[1] got overwritten with r[0] somehow...

If I add

<http://pulumi.log.info|pulumi.log.info>(f"[DEBUG] regions = {regions}")

I see what I expect, e.g.

[DEBUG] regions = ['us-east-2', 'us-west-2']

so that appears to be working fine.

Yep. I'm assuming it's a Pulumi bug then. I'm going to try to reproduce

For now I just updated the code in the AWS dashboard, but that makes me feel uneasy 😬

Bugs get filed in the appropriate git repo. I don't know if this is an AWS or a core Pulumi issue.. mabye @modern-spring-15520 would be able to direct you?

https://github.com/pulumi/pulumi/issues should work, with as much details to repro as you've got.

Well… I put together a simple proof of concept trying to reproduce what I’m seeing, but I’m not able to reproduce it either 🤔 #sanity-check

So it’s working on the simple example?

Yeah. I’m gonna pour coffee on my head and see if I missed something in my main app/stack

I’m gonna pour coffee on my head

I think you might be using the coffee wrong 🙂

seems to be working everywhere now. Lambda source code is updating as expected. Pour coffee on the keyboard?

Probably threatening with coffee is enough to scare it into proper action

weaponized beverages is the future of code