Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Handling of archive logs

Re: Handling of archive logs

From: <oratune_at_aol.com>
Date: Mon, 23 Oct 2000 21:46:20 GMT
Message-ID: <8t2bja$nqd$1@nnrp1.deja.com>

In article <39F4826C.43E02426_at_home.nl>,
  frank <fbortel_at_home.nl> wrote:
> Budgets are tight - as ever.
> Sites I have worked tend to have 1 or two days of uncompressed
 archived
> log files, and ditto compressed - thus having at least 2 copies on
 disk, and
> on 2 different tapes. Murphy, huh? Murphy was an optimist.
>
> Or -as one of the guys put it- tapes? a tape is just rust on wheels!
> And, indeed, I have witnessed the chaos when backups cannot be
> read from tape... But then again, no disaster plan testing...
>
> Frank
>
> "Howard J. Rogers" wrote:
>
> > The realist in me says if you think you can get away with it, why
 not?
> >
> > The purist in me says... the archive logs are the only thing
 standing
> > between you and data loss, so you play around with them at your
 peril.
> > Compression to me sounds like a very *bad* idea, since you are
 rather
> > relying on the compression working correctly, and the decompression
 to work
> > without a hitch when needed. Anything goes wrong at either end,
 and you've
> > just lost your logs, and hence the data they were supposed to be
 protecting.
> >
> > Managerially, it's also suspect: when the database goes down, your
 job is
> > supposed to be to get it back up and running again as quickly as
 humanly
> > possible. By compressing your logs, however, you've introduced
 another
> > stage in the process, and hence meantime to recover will inevitably
 be
> > longer.
> >
> > My suggestion? Buy some more hard disks.
> >
> > Regards
> > HJR
> >
> > "Stefan Fournier" <Stefan.Fournier_at_gmx.de> wrote in message
> > news:8FD3C9489StefanFourniergmxde_at_130.133.1.4...
> > > Hello,
> > >
> > > I have a question around the handling of archived logs and would
> > > like to hear how you do it.
> > > Today we did a import of a schema with about 500 MB. While the
 import
> > > ran, the database wrote so many archived logs that the filesystem
 went
> > > full.
> > > I then gzipped the archive logs to release diskspace and the thing
> > > finished ok.
> > > My idea now is to create a cronjob which zips new archive logs
> > > every 15 minutes or so.
> > > Is that reasonable?
> > >
> > > Thanks for any input.
> > >
> > > Regards,
> > > Stefan
> > >
> > >
> > > --
> > > Stefan.Fournier_at_gmx.de
>
>

This is quite true. I don't know how often I have tried to impress upon the on-site DBA and System Administrator to read-verify the backup tape prior to accepting it as valid. Simply because it "wrote" means absolutely nothing, as a successful write only indicates that the device was functioning properly. Read-verifying the backup tape ensures that the tape is usable (at least at the time the backup is created -- I won't go into the hazards and pitfalls of improper tape storage). I have seen many "backups" end up in the trash when the tape cannot be read at the crucial moment of recovery.

--
David Fitzjarrell
Oracle Certified DBA


Sent via Deja.com http://www.deja.com/
Before you buy.
Received on Mon Oct 23 2000 - 16:46:20 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US