Re: Help: STALE logs problem

[Snip]
> > The shutdown problem has since been resolved and it had nothing to do
> > with the rc scripts.
>
> I never said it did; I only commented that IF your rc scripts were
> properly configured you shouldn't have your shutdown problem, aside
> from a shutdown -r which does not invoke rc.
>
> I have never experienced an improper shutdown causing log files to go
> STALE (of course this is not to say it CAN'T happen). I HAVE, however,
> found errors in the LGWR trace files indicating that log files were not
> available (usually across a network). I have also found STALE log
> files when the database is not in archivelog mode. I have also seen
> this behaviour when the redo logs are excessively sized and, thus,
> never fill.
>
> From your statements I expected you to find errors associated with this
> behaviour, especially since the system was coming down and the databaSe
> wasn't. I am glad that you found no errors associated with this.
>

In point of fact, the V$s for the logs are distinctly dodgy in any event. For example: put database into noarchivelog mode. Make sure ARCH is off. Add a new log group. Query v$log -and the 'archived' column will say 'yes' when transparently that's impossible.

Add a new member to that new group, and query v$logfile. The new member will have a status of 'STALE'.

In fact, the status column crops up with 'stales' quite frequently, during normal use (and this is Oracle 8 and 8i on Solaris and NT).

I have my own rule of thumb: anytime I see a funny status, do a manual log switch through all groups, and re-query. If the status column clears, you know the previous status was one of those quirky 'features' of the V$. If it's still there, you know you have a problem!

Regards
HJR Received on Tue Nov 07 2000 - 00:49:22 CST