Re: Redo-Log Archiv Problem

"Peter Miller" <p.miller_at_brocom.de> wrote in message news:3969CA0A.62AA4949_at_brocom.de...
> Hello,
>
> I tried your hints and the real solution was to add 2 more redo log
> groups
> so there isn't a conflict between logging and archiving anymore. Only
> one little question: what's the deeper sin for creating e.g. 2 or 3
> members inbetween a log group (physically it's all the same: data-files)
> instead of
> building one or two more groups?
>

Not sure I entirely understand the question, but if I have, then you are demonstrating a bit of confusion between groups and members.

If you create multiple members for the same redo log group, then what you are in effect doing is mirroring that group -and hence ensuring that the failure of one mere member does not totally invalidate the entire group (the loss of entire group could well equate to lost committed transactions, if the group was current). Multiple members=redundancy.

Multiple groups, on the other hand, equate to nothing more than the ability to complete a checkpoint and/or an archive before switching back onto the original group.

Two entirely different things, and thus best not confuse them! The standard recommendation is three groups, mirrored twice (eg, a 3 x 3 matrix). Start adding more groups if (as seems has happened to you) you can't checkpoint quickly enough. Start adding more members if double redundancy doesn't suit your particular level of paranoia about always being able to recover the database.

Regards
HJR
> Thank you all for your help.
> Peter
>
>
> > >> Peter Miller <p.miller_at_brocom.de> schrieb:
> > >> >
> > >> > Thread 1 cannot allocate new log, sequence 4185
> > >> > Checkpoint not complete
> > >>
Received on Mon Jul 10 2000 - 00:00:00 CDT