On Wed, Sep 1, 2010 at 1:02 PM, Stephens, Chris <Chris.Stephens@xxxxxxx>wrote: > Currently 10.2.0.5 Linux_64. (I mentioned it in the original email. I > swear!) > > So you did, sorry, my mistake. > > > I believe the bug was discovered when we were on 10.2.0.3. > > The bug I am familiar with is from 9i - bug 2639679 Looking the doc for it it says it is an 8i bug, but in 9.2.0.8 it was still a problem for me, at least according to the procedure I wrote to deal with it. > > Fast refresh isn’t necessary to solve the immediate error condition but it > is necessary because the set of materialized view maintenance accounts for > >70% of the database I/O. There is no archiving strategy. We are > maintaining summary of large amounts of completely static data. The > database is in a data guard environment so it is also clogging up the > network. > > You may already know about this, but make sure you create an index on the SNAPTIME$$ column on the MV Log tables. I think this is still necessary in 10g. Failure to do so generates a lot of unnecessary IO. There's a support note about it somewhere, but I can't find it at the moment. > So, does anybody know what is up with dba_mview_keys? J > > Sorry, I've never looked at that issue. Jared Still Certifiable Oracle DBA and Part Time Perl Evangelist Oracle Blog: http://jkstill.blogspot.com Home Page: http://jaredstill.com