Project

General

Profile

Actions

Bug #40049

open

rgw: Abort Multipart Upload after X days.

Added by Manuel Rios almost 5 years ago. Updated almost 5 years ago.

Status:
Triaged
Priority:
Normal
Assignee:
Target version:
-
% Done:

0%

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

Description

Hi Ceph Devels,

Today we have found an unpleasant situation. We got reported to customer their use based in "radosgw-admin bucket stats --bucket = [bucketname]".

Ceph report us that customer is using 83TB in the full account.
Customer say that programs like CloudBerry Explorer with S3 Compatible reports just 20TB.

After checking the bucket it got thousand of incompleted multipart that appears to be the main discrepancy of the accounting.

For example, a bucket in the software report 3.69TB, and in CEPH report 8TB, near the double.

Why dont implement a global feature to gc level or lifecycleglobal level to autopurge the incomplete multiparts after X days?

https://gyazo.com/efed1a5e7ceec7a83d991c943dfa1a5f

https://gyazo.com/39040cd36b776e9ed856f574a9839aaa

As a workarround we can setup by a crontask to all bucket to inject a lifecyle policy but it not scalabled

https://docs.aws.amazon.com/AmazonS3/latest/dev/lifecycle-configuration-examples.html#lc-expire-mpu

We request to check if its posible to setup something like rgw_multipart_expire=7

Or show also incompleted multiparts on bucket size info when someone list externaly.

Actions #1

Updated by Greg Farnum almost 5 years ago

  • Project changed from Ceph to rgw
  • Category deleted (common)
Actions #2

Updated by Matt Benjamin almost 5 years ago

  • Status changed from New to Triaged
  • Assignee set to Matt Benjamin
Actions

Also available in: Atom PDF