REPL_INFO should not be used if slave was previously replicating properly
Opened this issue · 1 comments
dotmanila commented
REPL_INFO can be stale - for example, if slave drops from the network for some reason, the agent would reset replication to use the values from REPL_INFO which would be outdated. At worse this can break replication, lucky if the old binlogs does not exists anymore.
dotmanila commented
tags:#162653 CentOS 6.6, MySQL 5.5, pacemaker with corosync