Improper partition used when generating SNS Topic ARN

[farvour]

What happened?

In US Gov partitions, the provider generates the wrong ARN for the GetTopicAttributes call. This results in the pulumi_aws.sns.Topic() resource from failing to create when the lookup happens.

Example

When using a provider in the US Gov (or China I'm sure too) creation of the topic fails:

test_topic = aws.sns.Topic("test_tjf_1")

Output of pulumi about

pulumi about
CLI          
Version      3.112.0
Go Version   go1.22.1
Go Compiler  gc

Plugins
NAME        VERSION
aws         6.32.0
datadog     4.28.0
kubernetes  4.11.0
postgresql  3.11.0
python      unknown
random      4.16.1

Host     
OS       darwin
Version  13.6.6
Arch     arm64

This project is written in python: executable='/Users/tomfarvour/.local/share/virtualenvs/pulumi-FBQqJsZg/bin/python3' version='3.11.7'

Additional context

This patch, which I presume patches the upstream TF provider, assumes it can use aws as the partition.

pulumi-aws/patches/0052-non-idempotent-sns-topic-creation.patch

Lines 23 to 28 in ba65249

	+func constructTopicArn(client *sns.Client, account, region, snsTopicName string) string {
	+ return fmt.Sprintf("arn:aws:sns:%s:%s:%s", region, account, snsTopicName)
	+}
	+
	+var snsGlobalMutex sync.Map
	+

Contributing

Vote on this issue by adding a 👍 reaction.
To contribute a fix for this issue, leave a comment (and link to your pull request, if you've opened one already).

[t0yv0]

It could be the case that our patch introduced a new failure mode:

	if !tfresource.NotFound(err) {
+		return diag.FromErr(err)
+	}
+

If we're not always predicting ARN correctly, this could surface a failure from by-ARN locator that wasn't a failure in upstream provider.