Re: Problems with Advanced Replication...

HI,

Me too 6 months with 8.0.5, only to be told of a bug that prevented it from working. Now I'm on 8i and its a lot better.

there are a number of points to add
When getting ready for replication,

Create a separate table space for the replication Data dict objects.

do this before runnig the catproc script.

 size of above file about 300mb, initial storage 512k next 1m 1% increase.
alter system default tablespace to the above created table.

then run the catproc script.

check for errors, I have found on upgrades from 8.0.5 and re-installations of 8I, I can never get clean packages for sys or system usually I get 1 or 2 invalids.

Even re-compiles will not fix them.

finally note that the replication code packages oracle creates , go into the schema that holds the tables, so make sure there is plenty of space.

2. When adding a new master group and adding object to it, do not select a destination site, just select the items, generate replication support (minus the dest sites).
(doing it this way, if you get any errors you only have to clean up the master site, if you get a generation error and have dest sites setup things can get very messy, and in some cases to clean up the schemas you need to call oracle tech support.

once this has finished and the state is Quiesed, go back and add the dest sites, then resume replication.

be prepared to wait for several hours for the state to change from quesed to normal if the table is large and you have a slow link.

3.once the basics are working MAKE A BACKUP at all sites involved. 3a. also you must have an iron clad backup system at ALL SITES that are replicated, if possible, using hot backup & re-do logs.

4. watch out for firewalls.

5. you'd also better ensure you have conflict resolution, pls be aware that you CANNOT just take an oracle database & code, and replicate it. The code/application must take into account the replicated environment, ESP.pre-insert triggers used to populate key fields.

6. watch out for pushing transactions outside of a current time zone, ESP. if you use time stamps as a conflict resolution method.

I assume you are aware of some/most of these gotya's

steve

Roger <NOrgSPAM_at_mcs-hh.de> wrote:

> Hi,
>
> These are the key points I've found through trial and mostly error while
> struggling with replication on 8i. I've got it working now but it takes
> some patience and understanding of the concepts involved:
> ------------
> Thanks very much, after some searching on metalink I found a document that
> covered some of your points. So points 1-5 where already applied without
> success.
>
> >>6. Once you get the environment setup and are waiting for dba_repcatlog to
> clear out, try pushing the transactions by using
> DBMS_REPCAT.DO_DEFERRED_REPCAT_ADMIN.<<
> -----------------------------------
> How long can this take ? I must admit that I don't yet understand the
> concept of dba_repcatlog. What happens when dba_repcatalog 'clears out' ?
> AFAIK dba_repcatlog is a queue for replication catalog changes, is this
> right ?
>
> >>Don't be afraid to start all
> ---------------------------------------
> I think this is the first thing I will do in the morning: Recreate all
> instances an start from the beginning.
>
> You aren't working on a production machine, are you? <<
> ------------------------------------
> Not yet but I have to in about two weeks....
>
> Thanks very much for your help
> Roger
>
>
> Wayne Menzie <waynem_at_bosmedtechdotcom.nospam> schrieb in im Newsbeitrag:
> 8F12BA272wayneshammalammading_at_206.66.240.253...
> >
Received on Tue Apr 11 2000 - 00:00:00 CDT