damccorm opened a new issue, #20841: URL: https://github.com/apache/beam/issues/20841
If unspecified by user, source creates the temp dataset ID here: [https://github.com/apache/beam/blob/release-2.27.0/sdks/python/apache_beam/io/gcp/bigquery.py#L786](https://github.com/apache/beam/blob/release-2.27.0/sdks/python/apache_beam/io/gcp/bigquery.py#L786) This means that re-runs of same source split workitem will have different temporary dataset IDs. If some split() calls fail before cleaning up the dataset, such datasets will not be cleaned up even if the job is successful (after workitem retries). Instead of doing this, we should create a temp dataset ID at source creation so that it is shared between re-runs of the same workitem. This might be incompatible with templates so we might have to wait till we have a SDF-based BigQuery source. (Java potentially have a similar bug in BigQuerySource but we have a withTemplateCompatability() option for Java which runs BigQuery using DoFns which should perform the cleanup correctly) Imported from Jira [BEAM-12002](https://issues.apache.org/jira/browse/BEAM-12002). Original Jira may contain additional context. Reported by: chamikara. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
