Configuration considerations for Server Reassignment
Prior to v6.1.2.1, when monitored servers were reassigned, typically because the CEP ascertained a previously running EP has stopped, or a server rebalance was initiated, they were reassigned one at a time, serially.
Each NodeService
thread was stopped on its previous active EP, if the EP was
still running, and then restarted on its new active EP. If you’re running v6.1.2.1, or later, server
reassignments will be done, by default, in parallel, which speeds up the reassignment process
significantly.
This is controlled by two engine properties:
MULTI_THREAD_SERVER_REASSIGNMENT
Default value: True
SERVER_REASSIGNMENT_QUEUE_SIZE
Default value: 100000
SERVER_REASSIGNMENT_THREADS
Default value: Number of CPUs IBM® Sterling Control Center Monitor ascertains the server it is running on has
For faster server reassignment processing, it is not advisable to change these default values.