Networker

Re: [Networker] Staging in Networker

2002-11-04 15:56:11
Subject: Re: [Networker] Staging in Networker
From: "Thomas, Calvin" <calvin.thomas AT NACALOGISTICS DOT COM>
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Date: Mon, 4 Nov 2002 12:55:56 -0800
Yes, it is working for me on 6.1.1. I have verified it in the media database
using a 78GB file spanned across 3 - 35GB drives, but I haven't recovered it
from tape yet to be 100% sure.

I first setup 3 partitions, and only put 2 in the staging policy.  Networker
staged the parts from the 2 drives in the staging policy, could not stage
the rest, and marked the entire file as "invalid ss" on the third partition.


Here is how it works now.
When a save set spans from one file device to a second, staging pulls the
complete save set from "both" devices as long as both devices are included
in the staging policy.
The sequence of events is like this.
1. Networker pulls the first half from the first device.
2. Then it pulls the second half from second device.
3. Then it deletes the Save Set number from the media index.
4. The original save sets from 1. above then show up as "invalid ss"
(because the ss has been deleted from the media database)
5. Then Networker comes back and uses the "recover space" setting (Find it
in NwAdmin, Stage, View Details) to delete the "invalid ss" save sets that
show up.
This is why both devices must be in the staging policy for this to work.


Another interesting item:
When the save set is transfering to tape, it shows up in the media data base
for the tape as "aborted".  When the save set is finished, it changes from
"aborted" to "browsable".  In case anything interupts the clone, the save
set is already marked with the appropriate status.  I have watched the media
database while the save sets are transfered over one by one, and I am quite
pleased at this point....


Calvin E. Thomas
UNIX System Administrator
NACA Logistics





-----Original Message-----
From: Glenn Britton [mailto:britton AT NOC.UTORONTO DOT CA]
Sent: Thursday, October 31, 2002 10:25 AM
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Subject: [Networker] Staging in Networker


Calvin,
I am interested in your points #3 and #6.  I opened a case with
Legato Tech. Support in December 2000 (case #2053912) and
they concluded that staging policies could not handle savesets
which spanned file devices.  I was running Networker 6.0.1 build 174
on Solaris 6 or 7 at the time and had two partitions on separate spindles
 and two Networker file devices as you describe, but any saveset that
spanned 2  file devices was never staged to tape.
nsrstage uses nsrclone, and cloning savesets
that spanned tapes never seems to be a problem for us.
The man page for nsrclone in the EXAMPLES section states:
"Only complete savesets can be copied by nsrclone".

The explanation I was given was that staging was somehow marking savesets
as incomplete if they spanned file devices and that staging was
not designed for the way I wanted to implement it.   :(

It would be great if this limitation has been fixed since 6.0.1 or
if Legato Technical Support gave me an incomplete explanation 2 years
ago.

Glenn Britton
University Of Toronto
Computing and Networking Services
255 Huron Street
Toronto, ON, CANADA
britton AT noc.utoronto DOT ca
>
> Date:    Wed, 30 Oct 2002 12:35:15 -0800
> From:    Calvin Thomas <calvin.thomas AT NACALOGISTICS DOT COM>
> Subject: Re: Staging in Networker?
>
> I am currently working on this.
> I asked the same question(s) previously and got some good info from this
> group. Here is what I have found out so far.
>
> 1. Do NOT use a single file device. If you do, this is what happens.  When
> the file device is full, it will stop the backup, and stage off to tape.
If
> you have a single large file device this will take a long time, and your
> backup job will fail.  Not Acceptable.
> 2. Create 2 or more partitions, and a file device in each.  I am setting
up
> 5 - 35 GB devices as I write this.  Then you total up the amount of space
> that will give you (175GB in my case),  figure out the largest backup you
> will need to do (78 GB in my case), and set the high water mark of all the
> devices to ensure that you always have enough space for your largest
backup
> ( 80GB for me).
> 3. When you backup to one file device and the file device gets full, it is
> marked read only, and the save job continues on the next file device.
> 4. When one file device is stageing off to tape, a different file device
is
> accepting data for your backup job.
> 5. A file device can either Read, or Write, but NOT both at the same time.
> Hence the need for multiple devices.
> 6. When you set up a staging policy in networker, be SURE and include all
> the file devices in your pool in the same stageing policy.  This insures
the
> staging policy can transfer the complete save sets from hard drive to tape
> without leaving part of a save set on a hard drive that staging can't
> control.
> ....
> ....
>
> Calvin E. Thomas
> UNIX System Administrator
> NACA Logistics
>
>

--
Note: To sign off this list, send a "signoff" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=

--
Note: To sign off this list, send a "signoff" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=

<Prev in Thread] Current Thread [Next in Thread>
  • Re: [Networker] Staging in Networker, Thomas, Calvin <=