Veritas-bu

Re: [Veritas-bu] Failover Storage Groups

2007-06-06 14:59:22
Subject: Re: [Veritas-bu] Failover Storage Groups
From: "Martin, Jonathan" <JMARTI05 AT intersil DOT com>
To: "Brooks, Jason" <brooksje AT longwood DOT edu>, <veritas-bu AT mailman.eng.auburn DOT edu>
Date: Wed, 6 Jun 2007 13:49:46 -0400
Unfortunately that's not how it works (if only /sigh.)

Failover works as follows - Use the first storage unit in the list that
is not full or down (Failover). If the storage unit is only busy, the
policy waits to write to it. The other storage units will be used as
failovers. 

So if you run a job, and it fills up DSSU1 and fails, that job fails.
There's no job logic to make the next job go to the next DSSU or DISK.
The job retries, sees that DSSU1 is not full now, and uses that one
again (unless another job has filled it up.)

Best bets are to 1) Manage which data gets written where and have lots
of extra space available.  2) Manage your high water mark so if there's
not enough space free to write your biggest job the DSSU is "full" /
needs to clear out.  As far as destaging, that's just complexity added
to the mix and I wouldn't recommend destaging to tape while writing high
speed backups.

-Jonathan

-----Original Message-----
From: veritas-bu-bounces AT mailman.eng.auburn DOT edu
[mailto:veritas-bu-bounces AT mailman.eng.auburn DOT edu] On Behalf Of Brooks,
Jason
Sent: Wednesday, June 06, 2007 1:29 PM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] Failover Storage Groups

Environment: NBU 6.0MP4 + engineering binaries Windows 2K3 Master, Linux
RHEL3 Media SAN connected, DSSU drives Quantum/ADIC i2K

I've recently been fighting with 129s on our weekend fulls, particularly
with some Linux hosts, due to scheduling.  This weekend, one host failed
to backup several mount points with a 129, DSSU Full.  Evidently, by the
Activity Monitor, no destaging initiated.  

In the Storage Group for this schedule, I have a DSSU first, then a tape
drive configured as failover.  Why didn't the jobs drop to tape?  Or
does the term failover here not reflect dynamic, mid-job failover?  Each
of the jobs that failed has written data.

Thanks,
Jason

------------
Jason Brooks
Computer Systems Engineer
IITS - Longwood University
voice - (434) 395-2034
fax - (434) 395-2035
mailto:<brooksje AT longwood DOT edu> 

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

<Prev in Thread] Current Thread [Next in Thread>