Project

General

Profile

Bug #53512

radosgw-admin bucket rm --bucket=${bucket} --bypass-gc --purge-objects failing crashing on buckets having incomplete multiparts

Added by Pritha Srivastava about 1 year ago. Updated 6 months ago.

Status:
Pending Backport
Priority:
High
Target version:
-
% Done:

0%

Source:
Tags:
gc backport_processed
Backport:
pacific octopus
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

radosgw-admin bucket rm --bucket=testbucket1 --bypass-gc --purge-objects crashes

Steps to Reproduce:
1. Create a bucket :
s3cmd mb s3://testbucket1

2. Abort the multipart upload :

s3cmd put /boot/initramfs-0-rescue-6d298da93b6e47828ddf0d55f92b4461.img s3://testbucket1
upload: '/boot/initramfs-0-rescue-6d298da93b6e47828ddf0d55f92b4461.img' -> 's3://testbucket1/initramfs-0-rescue-6d298da93b6e47828ddf0d55f92b4461.img' [part 1 of 6, 15MB] [1 of 1]
15728640 of 15728640 100% in 0s 21.19 MB/s done
upload: '/boot/initramfs-0-rescue-6d298da93b6e47828ddf0d55f92b4461.img' -> 's3://testbucket1/initramfs-0-rescue-6d298da93b6e47828ddf0d55f92b4461.img' [part 2 of 6, 15MB] [1 of 1]
15728640 of 15728640 100% in 0s 22.57 MB/s done
upload: '/boot/initramfs-0-rescue-6d298da93b6e47828ddf0d55f92b4461.img' -> 's3://testbucket1/initramfs-0-rescue-6d298da93b6e47828ddf0d55f92b4461.img' [part 3 of 6, 15MB] [1 of 1]
65536 of 15728640 0% in 0s 1301.21 kB/s^CERROR:
Upload of '/boot/initramfs-0-rescue-6d298da93b6e47828ddf0d55f92b4461.img' part 3 failed. Use
/usr/bin/s3cmd abortmp s3://testbucket1/initramfs-0-rescue-6d298da93b6e47828ddf0d55f92b4461.img 2~eltPdbtxJhLeQmK2d0uvgBeXjS6kDvB
to abort the upload, or
/usr/bin/s3cmd --upload-id 2~eltPdbtxJhLeQmK2d0uvgBeXjS6kDvB put ...
to continue the upload.
See ya!

3. Run the following command to remove the bucket :
  1. radosgw-admin bucket rm --bucket=testbucket1 --bypass-gc --purge-objects

Related issues

Copied to rgw - Backport #53563: octopus: radosgw-admin bucket rm --bucket=${bucket} --bypass-gc --purge-objects failing crashing on buckets having incomplete multiparts New
Copied to rgw - Backport #53564: pacific: radosgw-admin bucket rm --bucket=${bucket} --bypass-gc --purge-objects failing crashing on buckets having incomplete multiparts New

History

#1 Updated by Casey Bodley about 1 year ago

  • Status changed from New to Pending Backport
  • Tags set to gc
  • Backport changed from pacific,octopus,nautilus to pacific octopus

#2 Updated by Backport Bot about 1 year ago

  • Copied to Backport #53563: octopus: radosgw-admin bucket rm --bucket=${bucket} --bypass-gc --purge-objects failing crashing on buckets having incomplete multiparts added

#3 Updated by Backport Bot about 1 year ago

  • Copied to Backport #53564: pacific: radosgw-admin bucket rm --bucket=${bucket} --bypass-gc --purge-objects failing crashing on buckets having incomplete multiparts added

#4 Updated by Cory Snyder about 1 year ago

The RadosMultipartUpload::abort method that was updated in the associated PR never made it to Pacific or Octopus. (https://github.com/ceph/ceph/pull/42266 was not back-ported). Do we need to backport that PR first or should the backports of this issue be cancelled?

#5 Updated by Casey Bodley 11 months ago

we shouldn't need to backport the zipper changes for RadosMultipartUpload::abort, no. in pacific/octopus, this logic is in rgw_multi.cc:abort_multipart_upload() at https://github.com/ceph/ceph/blob/pacific/src/rgw/rgw_multi.cc#L260-L269

it looks like the change from https://github.com/ceph/ceph/pull/44237 would apply cleanly to that function instead

#6 Updated by Backport Bot 6 months ago

  • Tags changed from gc to gc backport_processed

Also available in: Atom PDF