I would like your feedback as for the best way to fix this bug.
Our documented online GTID deployment procedure (https://www.percona.com/doc/percona-server/5.6/flexibility/online_gtid_deployment.html) never permits local workload on a node with gtid_deployment_step = ON. On slaves, when it is enabled, only replicated master workload is running, and the variable is reset to OFF on the new master promotion.
Based on this, it seems that it makes sense to implement behaving as if under (super-)read-only when gtid_deployment_step is set, under this bug, in addition to other fixes.
Do you agree or am I missing some intended use scenarios?
I would like your feedback as for the best way to fix this bug.
Our documented online GTID deployment procedure (https:/ /www.percona. com/doc/ percona- server/ 5.6/flexibility /online_ gtid_deployment .html) never permits local workload on a node with gtid_deployment _step = ON. On slaves, when it is enabled, only replicated master workload is running, and the variable is reset to OFF on the new master promotion.
Based on this, it seems that it makes sense to implement behaving as if under (super-)read-only when gtid_deployment _step is set, under this bug, in addition to other fixes.
Do you agree or am I missing some intended use scenarios?