Veritas-bu

[Veritas-bu] Sporadic, if not erratic NB database backups

2001-12-26 19:52:23
Subject: [Veritas-bu] Sporadic, if not erratic NB database backups
From: MarelP AT AUSTRALIA.Stortek DOT com (Marelas, Peter)
Date: Thu, 27 Dec 2001 11:52:23 +1100
How does cron know when Netbackup is inactive?
I thought the requirement was for Netbackup to be
inactive while catalog backups are taken, hence why Netbackup
schedules them.

Regards
Peter Marelas

-----Original Message-----
From: Peter L. Buschman [mailto:plb AT iotk DOT com]
Sent: Saturday, 22 December 2001 7:22 AM
To: Winkeler, Paul; 'Greg Nelson'; Winkeler, Paul
Cc: veritas-bu AT mailman.eng.auburn DOT edu
Subject: RE: [Veritas-bu] Sporadic, if not erratic NB database backups



Paul,

I'd recommend disabling automatic backups from within the NetBackup 
scheduler completely.  For an active system
like yours, neither of the two options, performing a database backup after 
a successful schedule or backup, is going
to be effective.  On such a system, you not only run the risk of not 
getting any database backups like you are now,
but possibly getting them too frequently.  In the latter case, you can end 
up with both of your database backup tapes
overwritten in a matter of hours, something that can be disastrous if it 
happens following a corruption event.

Automatic backups can also leave your system unable to run jobs for large 
blocks of time if the indexes are large.  I've
seen some systems that were running an hourly schedule and backing up the 
databases after every successful schedule.
On this system, it took half an hour for the database backup to complete, 
so for 30 minutes out of every hour, no backups
could run.  This eventually caused lots of hourly backups to miss, not to 
mention wearing out the database backup tapes
by writing to each one 12 times every single day.

I nearly always disable the automatic database backups and instead schedule 
bpbackupdb to run twice a day from cron.
This way, I always have two valid database copies within the past 24 
hours.  With scripting, it is also possible to get around
NetBackup's two media ids for database backups limitation as well.

Regards,

--PLB

At 01:24 PM 12/21/2001 -0500, Winkeler, Paul wrote:
>Greg
>
>That sounds like it might be the cause of the problem;  Our backup schedule
>is pretty much active 24 hours a day, 7 days a week.  At night we run the
>usual plethora of system backups but in the wee hours of the morning we
>split off a mirror of a 2.5Tb database and its backup takes the better part
>of the day (mixed with steady stream of Oracle archive logs) until the
whole
>process repeats itself come evening time.
>To eliminate further confusion as to how the NB database backup is
>configured, here is the output of bpsyncinfo:
>--------------------------------------------------------------
>Frequency of DB Backup:  after each successful backup schedule
>
>   Server: bkupsrv1.bkup
>   Sequence # 1    Last Media Used: S00413
>
>      Written            Allocated          Type   Density  Media
>      -----------------  -----------------  ----   -------  -----
>   1  12/09/01 12:59:07  02/12/01 10:28:28  RMedia dlt      S00413
>   2  12/06/01 12:36:07  12/04/01 17:19:18  RMedia dlt      S03640
>
>   Paths Included:
>     opscenter.bkup:/usr/openv/netbackup/db
>     opscenter.bkup:/usr/openv/volmgr/database
>     bkupsrv1.bkup:/usr/openv/netbackup/db
>     bkupsrv1.bkup:/usr/openv/volmgr/database
>     bkupsrv2.bkup:/usr/openv/netbackup/db
>     bkupsrv2.bkup:/usr/openv/volmgr/database
>--------------------------------------------------------------
>The DBBACKUP_CALLED log file has these last few lines:
>--------------------------------------------------------------
>Sun Dec 9 13:29:31 EST 2001 NetBackup databases backed up on TAPE S00413
>Sun Dec 9 13:37:31 EST 2001 NetBackup database backup FAILED to TAPE S03640
>Mon Dec 10 23:01:37 EST 2001 NetBackup database backup FAILED to TAPE
S03640
>Mon Dec 10 23:03:17 EST 2001 NetBackup database backup FAILED to TAPE
S03640
>Tue Dec 11 18:43:14 EST 2001 NetBackup database backup FAILED to TAPE
S03640
>--------------------------------------------------------------
>indicating that there haven't even been attempts to do a backup since last
>December 11th!
>
>-----Original Message-----
>From: Greg Nelson [mailto:greg.nelson AT veritas DOT com]
>Sent: Friday, December 21, 2001 12:57 PM
>To: 'Winkeler, Paul'
>Subject: RE: [Veritas-bu] Sporadic, if not erratic NB database backups
>
>
>Paul
>
>Make sure you don't have any ACTIVE backups that are hanging around for
>days.  If a scheduled backup completes but others are running then the
>catalog backup will not take place.  The catalog backup will only run when
>the NBU environment is inactive.
>
>-----Original Message-----
>From: Winkeler, Paul [mailto:PWinkeler AT officemax DOT com]
>Sent: Friday, December 21, 2001 11:36 AM
>To: veritas-bu AT mailman.eng.auburn DOT edu
>Subject: RE: [Veritas-bu] Sporadic, if not erratic NB database backups
>
>
>On any given day we typically run upwards of 450 jobs through the
scheduler;
>In spite of what we feel are high failure rates (roughly one every other
>day), the majority of the runs go through without a hitch.  The
>configuration tool states my NB database backup schedule options are:
>         1) Never (i.e. manual)
>         2) After each succesfull schedule
>         3) After each succesfull archive
>We currently use option 2.  But I am not seeing any NB backup activity for
>weeks on end!  How can this be?  What constitutes a successful schedule?
>
>-----Original Message-----
>From: Larry Kingery [mailto:larry.kingery AT veritas DOT com]
>Sent: Friday, December 21, 2001 12:16 PM
>To: Winkeler, Paul
>Cc: veritas-bu AT mailman.eng.auburn DOT edu
>Subject: Re: [Veritas-bu] Sporadic, if not erratic NB database backups
>
>
>A successful backup session means that at least one of the backups
>which ran was successful.  A single backup failure does not cause the
>NBU database to not get backed up.
>
>If all backups fail, the NBU db backup doesn't execute, which makes
>sense since there's effectively no new info in the db.
>
>Winkeler, Paul writes:
> >
> > We are set up to backup the NB database upon successful completion of
the
> > backup schedule.  Could it truly be the roughly one media error a day
that
> > causes us to barely do one NB database backup every week?
> >
> > If so, what *should* we set the NB backup schedule to?
> >
> > Getting even more nervous,
> >  Paul Winkeler
> > _______________________________________________
> > Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
> > http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
>
>--
>Larry Kingery
>                      Of couse, I could be wrong.
>_______________________________________________
>Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
>http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
>_______________________________________________
>Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
>http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu

_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu