RE: Save-while-active questions / concerns

Getting all of the users off the system should solve 90% of the problem.
Ending any user jobs that might update files should take care of the rest.
As long as nothing is being updated while the backup is syncing you should
have no problems.  Management Central needs to be shut down [ENDTCPSVR
SERVER(*MGTC)] while backing up QUSRSYS or you will get an object not saved
message.  If you have Apache web servers running you will need to end them
before the SAV or you will get object in use messages for the log files.

Getting a good checkpoint is not dependent on how few jobs are running, it
is dependent on the type of locks those jobs hold; you will not be able to
get a checkpoint on objects held by *EXCLRD, *EXCL or *SHRUPD locks.

Run your backups with LOG(4 00 *SECLVL) LOGCLPGM(*YES) and read the joblogs.
Any object not saved messages will identify the offending jobs.  I think
there is a TAATOOL that will scan them automagically if you have that.

Go over your restore strategy.  If you are trying to get a full backup
nightly using SWA it is critical that all objects get saved daily.  If you
do a full backup weekly or monthly and use SAVCHGOBJ daily you might not
care if an object that does not contain production data gets saved every
day.  I omit QPFRDATA from Daily and Weekly backups; it's almost 5GB in size
and I don't care if that library is not current in a DR situation.

You'll love SWA, on our prod system a full restricted backup runs about six
hours, a SAVCHGOBJ SAVACT(*SYNCLIB) syncs in about 6 minutes.

Regards,

Scott Ingvaldson
AS/400 System Administrator
GuideOne Insurance Group

"Remember, Only YOU Can Prevent Forest Fires!"
                                                -- Smokey Bear 

-----Original Message-----
date: Tue, 16 Dec 2003 14:49:57 -0500
from: "David A Parnin" <daparnin@xxxxxxxxxxxxxxxxxx>
subject: Save-while-active questions / concerns

Greetings everyone,

For our daily backups we have traditionally killed QINTER and another
subsystem or two, done the save's, then IPLed.  Awhile back I removed the
IPL and added code to restart what was shut down.  I am now getting ready
to implement save-while-active to reduce the downtime window even more.  I
don't have a problem with monitoring for when the checkpoint has been
established and restarting things but I do have a couple of questions.

How close to a restricted state does the system need to be to establish a
good checkpoint?  In the past even though QINTER was down there was still a
subsystem with some never-ending EDI jobs.  I assume that needs to be ended
to get a good checkpoint.  Perhaps a bigger concern is with ODBC jobs.  We
have a number of users that use Showcase queries.  They don't appear to
lock the files.  Should those be addressed?  I doubt that we've got anybody
using FPT or Telnet but should I end TCP/IP just to be safe?

I'm also concerned about the SAVACTWAIT parameter.  I want to avoid objects
being skipped.  Given that it has a 2 minute wait time per object if they
are in use this could add up if I happen to miss a couple of things.  In
case it matters we are at V5R1.  I was planning on using the *SYNCLIB
option.  I've read through the archives but didn't see these things
addressed.

Any thoughts would be appreciated.

Dave Parnin
Nishikawa Standard Company
Topeka, IN  46571
daparnin@xxxxxxxxxxxxxxxxxx
   
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.