hello, we are doing a redirected restore from a prod db to a test db...the problem is the prod db server has 16gb RAM and the test server has 2gb RAM...

we want the 50gb prod db put on the 50gb test db...the test server will have only 1 or 2 users to investigate issues, so performance is not an issue...

the prod db used the 'config advisor' and is sized to use 10gb out of 16gb - this sets several parameters with very high numbers of pages...

when we do the redirected restore, we get this message -
[IBM][CLI Driver] SQL1084C Shared memory segments cannot be allocated. SQLSTATE=57019

1. how to configure the target db to use lower config values on redirected restore?

2. or, how to restore from LARGE db config values to smaller config values?

3. or, how to configure a target db that does not yet exist, but will once the redirected restore is completed?

thanks for any insight!

I would guess that you do the restore, then before you use the restored DB, you change the values of the various config parameters and bufferpool sizes as needed to fit into the smaller space.

Andy

Try setting the DB2_OVERRIDE_BPF registry variable to a value that is suitable for the environment. This will activate the database with smaller bufferpools, not the ones defined in your production environment.

the DB2_OVERRIDE_BPF was a good idea but didnt work (prod bp is 256000 and set DB2_OVERRIDE_BPF to 12500)...

prod - 16gb - 4cpu:
instance_memory - auto(3600000)

db config:
appl_memory - auto(40000)
database_memory - auto(1250000)
dbheap - auto(25600)
pckcachesz = auto (96000)
bufferpool = 256000 (4k pages)


test - 2gb - 2 cpu:
instance_memory - auto(390000)

db config - how to set these lower or any other ideas to get to restore then change after redirected restore????

nevermind - just spoke with ibm tech support - they said it can't be done...

note: they are working on this ability to change db config parms in an upcoming release

Reduce the size of Production bufferpool before the backup and change them back after the backup is complete.

If the database is not too big use db2move to transfer the data and grab the objects (views , triggers , udfs etc) and create them in non production