Description of problem: Trino gains efficiency upon issuing multiple requests per single Query. the return results by RGW should be aligned with Trino expectations (otherwise queries are rejected or results are not accurate). upon aggregation statement (count) Trino pushes down a non aggregation statement, which retrieves an empty column. Trino issue parallel multiple s3select-requests, it seems that deviation in the result relates to the number of parallel requests. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
the following PR's deal with Trino / CEPH integration, and resolve various issues related to that integration. https://github.com/ceph/ceph/pull/49411 https://github.com/ceph/ceph/pull/50471 https://github.com/ceph/ceph/pull/52651
the Trino/CEPH integration is a title for "many things" the CSV flow is one integration point (out of several). meaning there are specific flows in RGW that handle the Trino/s3select/CSV integration (for one example ... splitting the object). the JSON and Parquet are the other integration points. and there could be more in the future.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Red Hat Ceph Storage 7.0 Bug Fix update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2023:7780