[Info-vax] Best way to stop batch queue startup when system is started

Stephen Hoffman seaohveh at hoffmanlabs.invalid
Mon Jun 20 13:24:41 EDT 2022


On 2022-06-20 17:13:36 +0000, Rich Jordan said:

> More fun with the DR server.  All the data and user disks are image 
> backed up and restored to the DR server each night (operations are 
> quiesced so we know they are clean).  System disk gets an occasional 
> standalone backup that is restored to the DR box production system disk 
> (it is running on an alternate disk for all of these ops), but most 
> nights an evil online backup is performed, transferred, and then only 
> specific files (SYSUAF, etc) are overlayed on the DR server's offline 
> production system disk.
> 
> Is there a best way to stop all batch queues when/if this system is 
> booted on its production disk?  It will have jobs that were pending at 
> the time backup was done (the standalone backups) which will of course 
> try to run when the batch queues start.  We want the job controller up, 
> queue manager _preferably_ up, print queues don't care, but batch 
> queues ALL stopped...

Stopping everything is easy. Restarting (just) what devices and batch 
queues you want restarted is harder, and requires a specific script, 
err, procedure for the restart.

I'd possibly look to (mis)use a site-specific tailored version of 
FIXQUE to transfer over the data in your home-grown "cluster" failover 
configuration; for the specific case here.  VSI has a copy of FIXQUE 
posted.

Pragmatically, you're writing migration and reinstallation tools. 
Piecemeal. You could step back from this piecemeal design approach, and 
stop trying to match the details of the environments (and those always 
get skewed), and work to migrate the data, and to manage and re-install 
the apps and tools tools using PCSI or VMSINSTAL or less-bueno 
home-grown installers. This also means an easier approach for creating 
and re-creating a test environment.

I'm not a big fan of using batch queues for job scheduling, either. 
Those tend to get tangled, which means incrementally creating a 
feature-deprived job scheduler on a foundation ill-suited for the goal. 
But you're already invested in this approach, so that's unlikely to 
change.


-- 
Pure Personal Opinion | HoffmanLabs LLC 




More information about the Info-vax mailing list