Re: REDO Logs - 3rd mirror?

"JustAnotherDBA" <jadba_at_bellsouth.net> wrote in message news:<sBpX9.58033$Jm2.54069_at_news.bellsouth.net>...
> UPS maintenance caused power to flicker off and on several times to our main
> Oracle production server causing the loss of 5 disk drives in 1 RAID array -
> 4 of which were used to hold just the REDO logs for 2 of 3 databases. These
> 2 Oracle databases were setup to multiplex on 2 RAID1 (mirrored) disks. This
> is a large Unix server, SunFire V880 8CPUs with 4 RAID A1000s (total disks
> about 54) and running Oracle8i 8.1.7.3 and 8.1.7.4 . The 3rd database is on
> totally separate RAID arrays with its own Oracle Home software and hardly
> missed a beat - 15 minutes to reboot.

<snip>
> Does anyone else even use a 3rd multiplex REDO member ? I haven't , but
> will probably do this especially considering the current 2 RAID1 mirrors are
> in 1 array.
>

HJR seems to think so (3 log members per redo log group) and posts as such quite frequently. I haven't been able to justify that configuration on w2k boxen yet ... but I do duplex.

It sounds like your company had much to lose in the redo logs that were lost.
For 8.1.7 I just use a dbms_job that calls a proc that switches the logfile every n minutes (e.g. 30) if it hasn't been switched since then.

Of course, the contents of the archived redo logs are just as vulnerable as the online redo logs if they haven't been copied/moved off of the same storage volume. But as you were able to perform media recovery, it seems that you had these covered on other volumes or offline.

UPS raid?

Paul Received on Sat Jan 25 2003 - 14:49:43 CST