By default an "abort cleanAllRUV" task will run forever (replica-certifyall: yes) until all the replicas can be reached and the clean task aborted. However, there is no way to abort the abort task. There should be a way to reissue a "new" abort task to kill the existing abort task. This can be done by checking if an abort task already is running for that rid, and simply remove the "certifyall" flag which will prevent the endless loop, and allow the abort task to finish.
Currently the only other option is to stop the server, edit the dse.ldif, and remove all the nsds5ReplicaAbortCleanRUV attributes.
Should this fix just go to 1.3.4? Or older version(s), too?
This is affecting many IPA customers, IMHO it should go into 1.2.11 and up, so it makes the next RHEL 6 release, etc.
Ticket has been cloned to Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1214074
33330c3..465e105 master -> master
Author: Mark Reynolds firstname.lastname@example.org
Date: Tue Apr 21 12:23:47 2015 -0400
Metadata Update from @nhosoi:
- Issue assigned to mreynolds
- Issue set to the milestone: 188.8.131.52
to comment on this ticket.