Casey Bodley

@theanalyst this looks like something we'll want to backport, could you please open a tracker issue?

``` 2022-09-02T04:22:12.684 INFO:teuthology.orchestra.run.smithi071.stderr:====================================================================== 2022-09-02T04:22:12.684 INFO:teuthology.orchestra.run.smithi071.stderr:FAIL: s3tests_boto3.functional.test_s3select.test_csv_parser 2022-09-02T04:22:12.684 INFO:teuthology.orchestra.run.smithi071.stderr:---------------------------------------------------------------------- 2022-09-02T04:22:12.685 INFO:teuthology.orchestra.run.smithi071.stderr:Traceback (most recent call last): 2022-09-02T04:22:12.685 INFO:teuthology.orchestra.run.smithi071.stderr: File "/home/ubuntu/cephtest/s3-tests-client.0/virtualenv/lib/python3.8/site-packages/nose/case.py", line 198, in runTest 2022-09-02T04:22:12.685 INFO:teuthology.orchestra.run.smithi071.stderr: self.test(*self.arg) 2022-09-02T04:22:12.685 INFO:teuthology.orchestra.run.smithi071.stderr: File "/home/ubuntu/cephtest/s3-tests-client.0/s3tests_boto3/functional/test_s3select.py", line...

full results in https://pulpito.ceph.com/cbodley-2022-09-01_15:28:29-rgw-pr-47848-distro-default-smithi

``` 2022-09-02T04:22:12.684 INFO:teuthology.orchestra.run.smithi071.stderr:FAIL: s3tests_boto3.functional.test_s3select.test_csv_parser ``` i think these are just coming from other subsuites like rgw/crypt and rgw/website that didn't change the s3tests branch in https://github.com/ceph/ceph/pull/47044, @alimaredia added a `qa/rgw/s3tests-branch.yaml`...

updated the suite-branch with https://github.com/cbodley/ceph/commit/e20bde8244d2a27c6431acd07e9967aaa8f04f01 and rescheduled in https://pulpito.ceph.com/cbodley-2022-09-02_14:33:28-rgw-pr-47848-distro-default-smithi/

> the other 20, do not seems related to s3select. agreed

i was able to reproduce one of the `to_string()` leaks. the [valgrind report](https://gist.github.com/cbodley/a631ce99a43637ca6966f1aa28af502a) has more debug symbols than the teuthology one, but still isn't all that helpful. it's pointing at...

WITH_BOOST_VALGRIND filters out the valgrind issues we were seeing around coroutines in https://tracker.ceph.com/issues/56500 the '--soname-synonyms=somalloc=\*tcmalloc\*' argument filters out lots of tcmalloc-specific issues like 'mismatched malloc/free'

@alimaredia would you be willing to coordinate the testing/merging of these branches?

@alimaredia seeing lots of failures there, are you sure you used the right suite-branch?