Managing Replication

I am looking for a tip on refreshing replication. When we refresh our test environment from prod, the replication environment becomes invalid. CDR is not able to initialize itself. This is because the names of our groups change, ie., g_prod - - i=10 becomes g_test - - i=10. Is it possible to simply do a dbexport syscdr -ss for both the primary and target servers (massage the *.unl files to reflect new names), drop syscdr, and then dbimport syscdr? Or do you have to go through the process of dropping replication entirely and rebuild each of the replicate sets? Environment: Solaris 2.6 IDS v7.30 UC6 Steve Romankiw

Anything is possible - as long as you know what to tweak. ;-) Seriously though --- It sounds like you are managing the test and productions instances as though it were a single enterprise. I'm guessing that's why you have different group names for your production and test instances. Why don't you simply treat these two environments as two seperate enterprises? Then all you would have to do is to maintain a seperate sqlhosts file - one for the production enterprise and one for the testing. That way you could easily refresh.... One warning however. When you run dbimport you are actually creating a new partition for the table that you are loading. Replication snooping relies on a flag being set in the table's partition page. This can be seen from oncheck -pt. If this flag is not set, then the row will not be snooped byddr_snoopy. Dbimport will not set this bit in the partition page. Therefor you will not replicate. That means that you can not use dbimport to refresh the databases in your test instances. Steve Romankiw wrote: > I am looking for a tip on refreshing replication. When we refresh our > test environment from prod, the replication environment becomes > invalid. CDR is not able to initialize itself. This is because the > names of our groups change, ie., g_prod - - i=10 becomes g_test - - > i=10. > > Is it possible to simply do a dbexport syscdr -ss for both the primary > and target servers (massage the *.unl files to reflect new names), drop > syscdr, and then dbimport syscdr? > > Or do you have to go through the process of dropping replication > entirely and rebuild each of the replicate sets? > > Environment: > Solaris 2.6 > IDS v7.30 UC6 > > Steve Romankiw