|
HP OpenVMS Systemsask the wizard |
|
The Question is: Hi, Execution queues which are autostart do failover to other nodes when autostart has been enabled. This is a good thing. But when the problem has been resolved and the initial host is again our friend, how do we tell only the ones which were still active at the time of the problem to haul ass back without dragging all possible queues back too? Thank-you. I can't seem to find where this is documented notwithstanding an extensive search. The Answer is :
That an active queue does not automatically return ("failover")
to a recently-rebooted (autostart) host is intentional behaviour.
This behaviour is intended to prevent the queue entries from being
unintentionally thrashed back to a host that is undergoing maintenance
or an upgrade. Manual intervention is required.
If you wish a queue to failover to (operate on) another node, you
must stop the queue and restart it with the desired list of nodes.
For example, if you wish to preserve existing jobs, you will want
to use a command sequence such as the following:
$ STOP/QUEUE/NEXT QUEUE$NAME
$ SHOW QUEUE/ALL QUEUE$NAME
$ START/QUEUE QUEUE$NAME -
/AUTOSTART_ON=(HOST1::TTA1:,HOST2::TTA2:)
The SHOW QUEUE command in the above sequence will need to be repeated
until all active jobs are completed. This DCL command sequence could
easily be replaced by a short DCL procedure which stops the queue,
loops pending completion of all active jobs, and then appropriately
restarts the queue.
If you can immediately stop the queue and are also able to abort all
active (executing) jobs, you can use the following command sequence:
$ STOP/QUEUE/RESET QUEUE$NAME
$ INITIALIZE/QUEUE QUEUE$NAME -
/AUTOSTART_ON=(HOST1::TTA1:,HOST2::TTA2:)
Existing discussions of the f$getqui lexical include the following
topics: (813), (1240), (2159), (3951), (4546), (4568), (4903), (5188)
(5471), (5567), (5651), (5793), (5982), (6315), (6877), etc.
|