vRA 7.1 - AWS Endpoint data-collection succeeds, but returns no compute resources.

[ITVictor]

I successfully setup an AWS Endpoint (Endpoint #1) several weeks ago and spun up a couple of EC2 instances to be sure it worked.  Then I deleted the endpoint and creds and associated Fabric group etc, and setup a new AWS Endpoint (Endpoint #2) to the same region, but a different AWS account.  It too works like a champ and can spin up EC2 instances.  I have now been tasked with recreating the initial AWS Endpoint #1 (2 separate AWS accounts same region).  I can create Endpoint #1 no problem and data-collection claims to succeed.  The first time I re-created EndPoint #1 I could see about half of the AWS Regions available as compute resources in the new Fabric Group for Endpoint #1, and all regions available for Endpoint #2.  I deleted the Endpoint #1, recreated it and now I see no AWS Regions available as compute resources at all for Endpoint #1.  Data-collection still claims to succeed.  There are no errors associated with data collection or the endpoint in the vRA logs (Log Viewer in vRA GUI).  Any ideas on how to force this endpoint to actually make AWS compute resources available would be most appreciated.

[ITVictor · Accepted solution]

The workflow to delete the Original AWS endpoint crashed and left an old (still functional) credential in the SQL database (DynamicOps.AmazonWSModel.AccessKeys table).  According to engineering this is a known bug. Apparently, each time we tried to create the new endpoint and specify new credentials, vRA 'saw' the old credentials, and would therefore not reconize data collection as new information and would not populate any new Compute Resources.  To fix the issue, we modified the old credential entry to 'break it' with a bogus access key and then it accepted the new credentials and worked properly and displayed the appropriate compute resources.