Project

General

Profile

Actions

Bug #43357

open

multisite: automatically detect radosgw-admin 'sync init' commands

Added by Casey Bodley over 4 years ago. Updated 12 months ago.

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

0%

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

Description

radosgw-admin provides commands to clear the sync status for metadata, data, and bucket sync. the intent is to force radosgw to restart a full sync

however, radosgw doesn't notice this empty sync status (at least for metadata and data sync) until the gateway is restarted. and, if incremental sync is already in progress, it may overwrite the sync status and avoid triggering a full sync altogether

so where possible, changes to these sync status objects should be guarded by cls_version to detect racing writes; and when multisite sync is trying to write out an updated marker but gets notified of a race via ECANCELED, it should back out and reread the sync status


Related issues 1 (0 open1 closed)

Related to rgw - Bug #40177: multisite: delete full sync index before switching to incremental syncResolved

Actions
Actions #1

Updated by Casey Bodley over 4 years ago

  • Related to Bug #40177: multisite: delete full sync index before switching to incremental sync added
Actions #2

Updated by Casey Bodley over 4 years ago

  • Status changed from New to Triaged
Actions #3

Updated by Shilpa MJ almost 2 years ago

  • Assignee set to Shilpa MJ
Actions #4

Updated by Casey Bodley 12 months ago

  • Tags changed from multisite to multisite multisite-backlog
Actions

Also available in: Atom PDF