Personalize Recommender stuck PENDING or IN_PROGRESS but not viewable

0

I had a creation script fail during construction of some Recommenders and upon retrying it I'm finding a name collision. The problem is: I can't seem to get any evidence that the named resource actually exists (in any state).

In an attempt to delete (before re-creating) the Recommender, I did this (via CLI):

aws personalize delete-recommender --recommender-arn my-recommender-arn

That yielded an error:

An error occurred (ResourceInUseException) when calling the DeleteRecommender operation: recommender is in PENDING or IN_PROGRESS status: my-recommender-arn

Okay, fine. But I don't see it listed in the console anywhere ... let's check to see if exists:

aws personalize describe-recommender --recommender-arn my-recommender-arn

Which responded with:

An error occurred (ResourceNotFoundException) when calling the DescribeRecommender operation: The given recommender does not exist: my-recommender-arn

Huh. Maybe this 'in progress' recommenders aren't viewable? I have another being constructed right now, lemme check on that one:

aws personalize describe-recommender --recommender-arn my-other-recommender-arn

Which gave me:

{
    "recommender": {
        "recommenderArn": "my-recommender-arn",
        "datasetGroupArn": "my-dataset-group-arn",
        "name": "my-name",
        "recipeArn": "arn:aws:personalize:::recipe/aws-vod-more-like-x",
        "recommenderConfig": {
            "minRecommendationRequestsPerSecond": 1
        },
        "creationDateTime": "2023-09-13T17:00:07.459000+01:00",
        "lastUpdatedDateTime": "2023-09-13T17:00:07.783000+01:00",
        "status": "CREATE IN_PROGRESS"
    }
}

Uhhh, okay. It looks like IN_PROGRESS recommenders are visible.

So, it appears that I have a 'ghost' recommender that is neither visible nor deletable, but now blocks my progress because of a name collision.

This feels like a bug in the Personalize tooling? (The failed creation was hours ago on a small dataset, too, so I don't think this is just a case of eventual consistency of ARNs through the AWS system.)

Murat
asked 7 months ago183 views
1 Answer
0
Accepted Answer

This 'undead'-like resource was eventually cleared, though it seems to have taken the better part of a full day. Still feels like a bit of a bug, minimally a pretty bad/broken UX that would likely also cause issues with any Cloudformation scripting -- i.e. erring because of a duplicated ARN but not actually being to describe that ARN.

In any case, for my specific ARN, this has now been resolved (by time).

Murat
answered 7 months ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions