All you really need is *LIB. I suggest you create a message queue to
get all the SWA messages, I usually call mine BACKUP and put it QGPL.
The command would look like:
The device could also be a save file if you use those instead of a tape
device. If a commit gets in the way after 120 seconds that object will
not be backed up and you'll see that in the job log and it will move on
to the next object. I'm assuming you do not have BRMS available to you
but that would make this easier with a one line control group. BRMS
would then handle the SWA processing.
I have update history set to no to keep the regular save processing in
charge of recoverability.
You can check the message queue to see how long the checkpoint takes to
get. If it's more than a minute then you might have some long lasting
Chief Technical Architect
Agile Technology Architects
On 3/2/2012 4:22 PM, Joe Pluta wrote:
I'm a neophyte when it comes to save while active. I want to save a
smallish production library (under 100GB) during production to replicate
it for testing on another machine. I want to do this daily during
offpeak (but still production) hours. I noticed a couple of options on
the SAVACT: *LIB vs *SYSDFN. Since I'm only saving one lib, *SYNCLIB
Anyway, does anyone have any recommendations for using save while
active? I realize this is what it's meant for, but that doesn't mean
there aren't considerations. I'm much more worried about minimizing
delays to production than I am about getting a 100% clean image for the
test machine. Because of that, I'm leaning towards
SAVACTWAIT(*NOCMTBDY) but I'd love to hear other people's experiences.