I have a CDK stack that contains a Timestream Database with a Timestream Table. There has been no changes to these since they were originally deployed.
When running the deployment locally, everything goes smoothly for both of the two accounts we have this stack in.
However, the CI/CD system is not allowed to deploy. It fails the same way for both accounts:
UPDATE_FAILED | AWS::Timestream::Table | <table name in stack> (<table name>) Resource handler returned message: "null" (RequestToken: 73f0b0b6-fee5-dba2-355f-d85a235942df, HandlerErrorCode: InternalFailure)
There is no error in CloudTrail. The only two visible events related to this are a CreateChangeSet and an ExecuteChangeSet. None of these show any errors.
I have checked the deployment template in S3. The template from the local deploy contains the RetentionProperties while the one from the CI/CD-deploy does not. The CDK defines these explicitly.
The Role used by the CI/CD system has these premissions on "*":
'timestream:DescribeDatabase',
'timestream:DescribeEndpoints',
'timestream:DescribeScheduledQuery',
'timestream:DescribeTable',
'timestream:ListDatabases',
'timestream:ListMeasures',
'timestream:ListTables',
'timestream:UpdateDatabase',
'timestream:UpdateTable',
'timestream:UpdateDatabase',
'timestream:UpdateTable'
I have tried giving the deployment role all rights with 'timestream:*' and the result is the same.