On 25-Aug-2014 15:50 -0500, Stone, Joel wrote:
Each evening we bring down all users and run save-while-active on
production files.

Once in a while the save-while-active job cannot get a lock on a

After bringing down the online system, we would like to run a job
prior to the save that sends a text message to a few people showing
the JOB/lib/file of any production objects with a lock on them.

This would avoid the scenario where the SAVE job cannot get a lock on
each and every object for a moment.

Is a good method to <<SNIP>>

Seems there is a flaw in the attempt to use SWA. If the "online system" is _brought down_ [and "all users" as well], yet there are still locks on the objects of that application\system, then the problem would seem to be with if\how the application is being terminated to free those resources.

Having approached the issue from that perspective, there should remain no need to try to figure out who\what holds those resources still, except in what would hopefully be rare cases of defects in the processing that should both /bring down/ the application and prevent anyone starting any new instances of that application until after the checkpoint processing can be achieved.

This thread ...


Return to Archive home page | Return to MIDRANGE.COM home page