High Availability and HAWQ Ranger
This topic describes specific HAWQ Ranger failover scenarios and procedures. You must have registered and configured a standby HAWQ Ranger Plug-in Service as described in Register a Standby Ranger Plug-in Service for these failover scenarios to occur.
Should the HAWQ master node fail to communicate with the local Ranger Plug-in Service, the master automatically switches over to the Ranger Plug-in Service registered on the HAWQ master standby node. This operation should be transparent to all HAWQ users.
hawq state cluster command output identifies the master service state as
Down when the Ranger Plug-in Service has failed over to the master standby node:
20170515:01:15:26:045131 hawq_state:master:gpadmin-[INFO]:-- HAWQ master Ranger plugin service state = Down 20170515:01:15:27:045131 hawq_state:master:gpadmin-[INFO]:-- HAWQ standby Ranger plugin service state = Active
After failover to the standby Ranger Plug-in Service, the HAWQ master periodically attempts to re-establish contact with the service on the local node. The
hawq_rps_check_local_interval server configuration parameter identifies the polling time interval for this contact. When communication is restored with the Ranger Plug-in Service on the local node, the HAWQ master automatically switches back to the local service. This operation is similarly transparent to all HAWQ users.
If the HAWQ master node goes down, you will activate the master standby node, at which time the standby becomes the new HAWQ master. When the HAWQ master fails over in this manner, the master Ranger Plug-in Service also fails over to the standby node.
After activating a HAWQ standby master node, you must manually update the Ranger HAWQ service definition with the new HAWQ master node connection information. Update this information via the Ranger Admin UI.