Matt,
When Ik look at this snippet of your logfile:
Sun Aug 22 02:32:14 2004
Shutting down instance (abort)
License high water mark =3D 2
Instance terminated by USER, pid =3D 12911
lter tablespace RBS begin backup
Completed: alter tablespace RBS begin backup
Sun Aug 22 02:18:15 2004
I see that the Alter tablespace command isn 't written completely, the
'A ' is missing. To me it looks like the volume manager 's caches being
flushed at different times during the 'split brains ' period. Is that
possible, or every system just had another idea of what blocks to flush?
Or is it just a cut/paste error? Is it possible to find out whether the
actual shown 'lter tablespace ' just appears on a block split in the
file? Sorry, I 'm not a cluster filesystem specialist, just some guesses.
Best regards,
Carel-Jan Engel
===
If you think education is expensive, try ignorance. (Derek Bok)
===
On Wed, 2004-08-25 at 15:17, Adams, Matthew (GE Consumer & Industrial)
wrote:
> No idea at all. We do know for a certainty=20
> that the Veritas Cluster went "split brain "
> (i love that term our sysop came up with)
> and both sides of the cluster thought it owned
> the package in question and both sides mounted,
> opened and wrote to the database files.
>
> But since the clocks are in-sync I cannot come=20
> with a reason that the timestamps on the alert logs
> would be so out of order, even with both sides
> writing to it.
-- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
-- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ------
To unsubscribe send email to: oracle-l-request@(protected)
put 'unsubscribe ' in the subject line.
--
Archives are at http://www.freelists.org/archives/oracle-l/
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
-- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- --
