Networker

[Networker] Savegrp Parallelism - Funny Math?

2003-07-09 15:09:28
Subject: [Networker] Savegrp Parallelism - Funny Math?
From: "David E. Nelson" <david.nelson AT NI DOT COM>
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Date: Wed, 9 Jul 2003 14:09:21 -0500
Hi All,

Got a question here about savegrp parallelism and target sessions.

NW: 6.1.3
OS: Solaris 8
HW: E-450

I think of 'target sessions' as a "desired minimum". ie. I want AT LEAST X
sessions writing to this drive before I'll request another drive.

I think of 'Savegrp Parallelism' as an absolute max. ie. I want no more than X
sessions running under this group.

I think of 'Client Parallelism' as an absolute max. ie. I want no more than X
sessions originating from this client.

I think if 'Server Parallelism' as an absolute max. ie. I can not accept more
than X sessions total for the entire data zone.

Are the above statements correct?

If I BU'd one client (savegrp -v -c tbox-app Production) w/in a group
consisting of 45 clients using the following settings and h/w:

9840 Drives: 3 on the server; 2 on a SAN storage node
9840 Target Sessions: 4
Loaded/Mounted tapes in 9840 Drives: 0
Server Parallelism: 32
Savegrp Parallelism: 4
Client Parallelism: 4
Client Save Set: All
Client Filesystems: /, /nicapp, /boot, /dev/pts

BTW, 'tbox-app' is just a regular client - not a storage node of any type.

This is how NW responded:

Media Alert: Waiting for 4 tapes for pool 'Full'

** This is odd, it should know that only 3 drives can possibly be used, yet
it's requesting 4 tapes.

It subsequently loaded all three 9840's with tapes and wrote ONE savestream to
each drive.

** I expected NW to load one tape and write save What the heck is going on
here?  Why isn't 'target sessions' being used here?

There is no difference from above if I use 'savegrp -N 4 -v -c tbox-app
Production' (added '-N 4').  If I used '-N 2' two drives were loaded and
written to w/ ONE savestream to each drive.

Now, if any of the drives already had a volume loaded/mounted (no matter if
it's one drive or all 3 drives) and I issue 'savegrp -v -c tbox-app
Production', then all 4 streams go to ONE drive.  This is the expected
behavior.

Is this is known issue and is there a solution?

Thanks,
        /\/elson


--
~~ ** ~~  If you didn't learn anything when you broke it the 1st ~~ ** ~~
                        time, then break it again.

--
Note: To sign off this list, send a "signoff networker" 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>
  • [Networker] Savegrp Parallelism - Funny Math?, David E. Nelson <=