aws_s3_bucket_object eventual consistency error after object creation

[bpiper]

Community Note

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
• Please do not leave "+1" or "me too" comments, they generate extra noise for issue followers and do not help prioritize the request
• If you are interested in working on this issue or have submitted a pull request, please leave a comment

Terraform Version

0.10.8

Affected Resource(s)

• aws_s3_bucket_object

Expected Behavior

Resource creation successful.

Actual Behavior

aws_s3_bucket_object.foo: Failed to get object tags (bucket: some-bucket, key: /foo): NoSuchKey: The specified key does not

As is generally the case with eventual consistency issues, this only happens sometimes.

Steps to Reproduce

Just define a standard S3 bucket object like so and run apply:

resource "aws_s3_bucket_object" "foo" {
  bucket = "some-bucket"
  key    = "/foo"
  source = "${path.module}/foo"
  etag   = "${md5(file("${path.module}/foo"))}"
}

This seems like just another Terraform AWS eventual consistency issue. The resource is doing a HeadObject and a GetObjectTagging immediately after the PutObject. In this particular case that I encountered, it was the GetObjectTagging request that hit a 404. I have seen explicit retry logic used by other resources to accommodate eventual consistency, but there doesn't appear to be any in resource_aws_s3_bucket_object.go. Resources absolutely should be reading after write to ensure they are in a consistent state, but this needs to be done with some sort of retry/timeout logic.

[bflad]

The fix for this is merged and will release with version 1.56.0 of the Terraform AWS provider, likely in the next day or two. 👍

[bflad]

This has been released in version 1.56.0 of the AWS provider. Please see the Terraform documentation on provider versioning or reach out if you need any assistance upgrading.

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. Thanks!