RE: Save while active (SWA) question -- MIDRANGE-L

Jeff -

You will either want to use option one or submit both jobs to batch.  Be
sure that the monitor job starts before the backup job or you may miss the
CPI3712 or CPI3710 message, sometimes they can reach the sync point quickly.
Also note that using the RCVMSG command will cause your monitor job to be
perpetually in MSGW status, (Job ... not waiting for a specific message.)
this has been known to confuse operators who expect any job in MSGW status
to require intervention.

--------------------------
Evan -

If you use BRMS you should try using the MONSWABRM exit.  I have found this
to be very reliable.

Regards,

Scott Ingvaldson
iSeries System Administrator
GuideOne Insurance Group   


-----Original Message-----
date: Thu, 21 Oct 2004 23:14:25 +1300
from: Evan Harris <spanner@xxxxxxxxxx>
subject: Re: Save while active (SWA)  question

Hi Jeff

Why wouldn't you do them both in batch ? (just curious)

I don't really have a recommendation either way, or more exactly a good 
reason for why I'd probably go with Option 1.  I used BRMS for this which 
meant that option 1 seemed to be a more natural fit. Whichever way you go, 
build a timeout into your restart job to make sure it doesn't wait forever 
for the SWA checkpoint when the inevitable error comes along.

There are some operations that will still bite you once you have reached a 
checkpoint. For example, if you have any programs that save or restore or 
objects they will fail if they attempt to use the objects in the SWA 
library (at least in my  experience). I saw some stuff at one site where 
they transferred stuff around between machines using savf's and restored it 
when it arrived at the target system but this bombed during the SWA 
process. Kind of self evident that a library being saved wouldn't allow 
restores...

I'll follow this thread with interest.

Regards
Evan Harris


>I'm going to change most backups to use a SWA scenario.  In a SWA
>environment, you monitor a message queue for the checkpoint processing
>complete.  Is there a preference to one of the following ways of doing it:
>
>Option 1
>========
>Controlling CL program
>         - calls a CL to stop QINTER, jobqs, etc.
>         - submits a job to QCTL that monitors the msgq for checkpoint
>                 (when reached, this submitted job restarts QINTER, etc)
>         - performs the backup
>         - done
>
>Option 2
>========
>Controlling CL program
>         - calls a CL to stop QINTER, jobqs, etc.
>         - submits a job to QCTL that performs the backup
>         - monitors the msgq for checkpoint
>         - restarts QINTER, etc
>         - done
>
>In option 1, the SAVxxx commands are done interactive with the msgq
>monitoring in batch.
>
>In option 2, the msgq monitoring is done interactive with the SAVxxx
>commands in batch.
>
>Is there a caveat to one or the other of these?
>
>--
>Jeff Crosby
   
This message and accompanying documents are covered by the Electronic
Communications Privacy Act, 18 U.S.C. §§ 2510-2521, and contains information
intended for the specified individual(s) only. This information is
confidential. If you are not the intended recipient or an agent responsible
for delivering it to the intended recipient, you are hereby notified that
you have received this document in error and that any review, dissemination,
copying, or the taking of any action based on the contents of this
information is strictly prohibited. If you have received this communication
in error, please notify us immediately by e-mail, and delete the original
message.