aws_elasticache_cluster does not correctly handle deleted clusters

[apparentlymart]

If the cluster associated with an aws_elasticache_cluster resource is deleted outside of Terraform, a subsequent terraform refresh will fail:

Error refreshing state: 1 error(s) occurred:

* aws_elasticache_cluster.web: CacheClusterNotFound: CacheCluster not found: omt-jsmith-dev
    status code: 404, request id: (elided)

The expected behavior is that it would call d.SetId("") to tell Terraform that the resource no longer exists, so that plan can know to create a fresh instance of the resource.

In the code we can see that it doesn't have any special handling for the 404 case, and it looks like it was instead expecting to get a 200 OK response with an empty set of CacheClusters. However, even if an empty set had been returned I think this would have still behaved improperly since it doesn't call d.SetId("") even in that case.

[stack72]

@apparentlymart you are indeed correct , PR is on it's way. I just ran the following test:

1. Created a elasticache cluster as follows:

provider "aws" {
    region = "us-east-1"
}
resource "aws_security_group" "bar" {
    name = "tf-test-security-group"
    description = "tf-test-security-group-descr"
    ingress {
        from_port = -1
        to_port = -1
        protocol = "icmp"
        cidr_blocks = ["0.0.0.0/0"]
    }
}

resource "aws_elasticache_security_group" "bar" {
    name = "tf-test-security-group"
    description = "tf-test-security-group-descr"
    security_group_names = ["${aws_security_group.bar.name}"]
}

resource "aws_elasticache_cluster" "bar" {
    cluster_id = "tf-test-sdsdfsffwfqw"
    engine = "memcached"
    node_type = "cache.m1.small"
    num_cache_nodes = 2
    port = 11211
    parameter_group_name = "default.memcached1.4"
    security_group_names = ["${aws_elasticache_security_group.bar.name}"]
}

I then went to the AWS UI and destroyed it. When running terraform refresh, I got the following:

Error refreshing state: 1 error(s) occurred:

* aws_elasticache_cluster.bar: CacheClusterNotFound: CacheCluster not found: tf-test-sdsdfsffwfqw
    status code: 404, request id: eb7b246f-83c5-11e5-9d16-9150d7bfbe93

I changed the code of the Read method as follows:

    res, err := conn.DescribeCacheClusters(req)
    if err != nil {
        if eccErr, ok := err.(awserr.Error); ok && eccErr.Code() == "CacheClusterNotFound" {
            d.SetId("")
            return nil
        }

        return err
    }

I was then able to refresh the state as required

[apparentlymart]

Fixed in #3767!

[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 have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.