Is there anyone out there using HDR on v7.3x successfully?

I've got replication running, it seems to work great but create a temp table do a few bits and "crash, bang, wallop" replication failure. If you specifically use the with no log clause then it seems to fix it, but why is it a problem in the first place - temp tables don't replicate, right?

And another thing, I have a weekly batch process that always causes a dr ping timeout. the last stage of the process updates 250k records but not all in one statement. There is another way to write the script but it runs much slower that way and I have to guarantee batch process completion. Any Ideas?

We have used replication on 7.31 TC5 (Win NT) and 7.31ud4 (AIX) for a couple of years. To sart off with server1 is a standard server with DBSPACETEMP set to a dbspace called tmpdbspace. This has been created using the -t option on onspaces.

We have never successfully used filesytems a temporary space.

This server will be fine and anything that uses sort space (select * from "bigtable" sort by "indexed column") works fine. On the targetted secondary you must ensure DBSPACETEMP is set the same as the standard server.
We then archive the standard and restore on the targetted secondary then pair them up with onmode -d commands. The secondary on restoring the tape will report an invalid dbspace "tmpdbspace". We then delete this dbspace on the now primary, do a checkpoint with onmode -c which should completed ok - onstat -d reveals this dbspace has now disappeared. Then we create the tmpdbspace again and force a checkpoint with onmode -c - the temporary dbspace now appears on the secondary - onstat -d reveals it so.

Eh voila tmpdbspace exists and you should be able to use it without a problem.

Hope this helps.

Hi,

Temporary tables that are created in the temporary dbspace(s) do not replicate. Any other thing, like temporary tables created in the rootdbs or the database dbspace will be replicated. That's why the 'with no log' works in your case. Anyway it sounds like some bug. Have you checked with Informix Support?

And about your second question: it depends. If you can, try stored procedures, parallel jobs/sesions, play with the internal server parallelization (PDQPRIORITY), and so on.

Hope it helps