Bug 2021970

Summary: The rgw container crashes when trying to delete a bucket bypassing the GC and this bucket has an incomplete object
Product: [Red Hat Storage] Red Hat Ceph Storage Reporter: maugarci
Component: RGWAssignee: J. Eric Ivancich <ivancich>
Status: CLOSED DUPLICATE QA Contact: Vidushi Mishra <vimishra>
Severity: low Docs Contact:
Priority: unspecified    
Version: 4.2CC: cbodley, ceph-eng-bugs, gsitlani, kbader, maugarci, mbenjamin, tpetr, vereddy
Target Milestone: ---   
Target Release: 5.2   
Hardware: Unspecified   
OS: All   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2021-12-10 17:43:46 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description maugarci 2021-11-10 13:35:52 UTC
Description of problem:

The RGW container crashes when the following conditions are met:

1- Uploading of an object containing multiparts is interrupted from the client side.
2- The cluster administrator performs the removal of the objects in the buckets, where the incomplete object resides, bypassing the GC


Version-Release number of selected component (if applicable):

RHCS4.2

Container image: rhceph/rhceph-4-rhel8:4-57


How reproducible:

This can be reproduced as follows


Steps to Reproduce:
1. Uploading a file bigger than 15MB using s3cmd.
2. Interrupt the uploading process before it ends.
3. Try to remove the bucket where the file was uploading:
  `podman exec <RGW_container_name> radosgw-admin bucket rm --purge-objects --bypass-gc --bucket=tests-removing-objects --debug_ms=1 --debug_rgw=20`

Actual results:

The container crashes and the incomplete object cannot be deleted bypassing the GC.

Expected results:

The expected behavior is to be able to successfully remove the object bypassing the GC and this operation must not crash the container.

Additional info: