We were having issues while attempting to use the node-manager across two clusters. The first cluster would start up, properly pick master and correctly handle failover however for the second cluster a master would never be promoted and though failover seemed to work the cluster status was not being properly tracked by node-manager.
This fix has been applied to our staging environment and appears to be doing the right thing (now). In attempting to test this I found that there was going to be a lot involved and wanted to get feedback on what would make sense to test, at this point, if anything.
Insert the "root_znode" path before "master_redis_node_manager_lock" …
…and expose via accessor
This looks great! Thanks for the fix.
Glad to help.
+1 any chance of a gem release for this?
Just released redis_failover 1.0.2 with these changes. Thanks!