RE: Disaster recovery solutions

Rob,

I get the CPI1466 3 times every night during my full SWA backup. Job holds large number of locks.
Any library that has a large number of objects will cause this message.
QSYS- 27,232,
IC08XXPGMS- 22,432 our one 3rd party app, Netcracker/ICOMS program library

Paul

-----Original Message-----
From: MIDRANGE-L [mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of rob@xxxxxxxxx
Sent: Thursday, December 17, 2015 3:21 PM
To: Midrange Systems Technical Discussion
Subject: RE: Disaster recovery solutions

I think that telnet is rather a stateful setup.
Perhaps, with the right switching equipment, you can switch without even changing the IP address.
Look at it this way, you have RPG program ORDxyz running on PROD. It has a bunch of variables populated. Was written without anyone reading the 1960's (or 70's) era document (I've seen it) from IBM recommending you not chain with a record lock and then present it to the user but rather you reread it and update after the user enters to update. How can they jump from within the middle of a transaction in a system like that to a backup system?
Or you have this batch job running doing a cost roll up or some such thing. Takes an hour. Calculates a total from the whole run of this job.
Can you see that failing over seemlessly?

Mimix locks EVERYTHING on the backup system for no update outside of it.
You can read it. You can even stop this for awhile to do backups (we do).
Message ID . . . . . . : CPI1466 Severity . . . . . . . : 60
Message type . . . . . : Information
Date sent . . . . . . : 12/17/15 Time sent . . . . . . : 04:55:24


Message . . . . : Job holds large number of locks.
Cause . . . . . : The job 966040/MIMIXOWN/GDI_DBAPYA has 30000 locks.
30000
is a suspiciously large number of locks, and it indicates that an
application may be leaving locks behind.
Recovery . . . : Determine whether the locks are correct for the
application. To do this:

I have screens full of these messages...

Getting these locks takes a little time. Not too long really but not subsecond.


Rob Berendt