Description of problem: Behavior of max-saved-replicated-journal-size = 0 is not intuitive. It preserves infinite number of old replicas but it should preserves 0 old replicas. Steps to Reproduce: 1. Configure replicated HA topology with max-saved-replicated-journal-size = 0. 2. Do several failovers/failbacks. 3. Check whether the number of old replicas is 0 on backup node.
See also https://bugzilla.redhat.com/show_bug.cgi?id=1348322
The intention was that you should have at least 1 folder up. Otherwise the ID for the old folder would always be 1.
I have added this into upstream: https://github.com/apache/activemq-artemis/pull/758 it's an easy change. I will wait for further instructions on where and when to back port it.
when/where to back-port this? Do I have permissions?
The fix should go to the next CP release. So just add the commit to 6.4.z maintenance branch.
(In reply to Erich Duda from comment #5) > The fix should go to the next CP release. So just add the commit to 6.4.z > maintenance branch. This is something Brad will decide. Clebert, please hold on with backporting to your 6.4.z maintenance branch till the present BZ gets assigned to some specific CP.
PR sent https://github.com/hornetq/hornetq/pull/2115
Verified with EAP 6.4.16.CP.CR1
Released on June 20 2017 as part of the EAP 6.4.16 maintenance release.