George,
I'm late to this discussion, so forgive me if I'm repeating what
someone else said. With the examples you gave, I'd expect
the first command variation -- nsrjb -IEv -S 1-6 -- to run as
many mount/load/verify/unload operations in parallel as is
specified by the max parallelism value, and I'd expect the
second variation -- nsrjb -IEv -S 1-3 -f /dev/rmt/0cbn plus
nsrjb -IEv -S 4-6 -f /dev/rmt/1cbn -- to run as if the value of
max parallelism is set to 2. I forget the exact reason why,
but with each of my jukeboxes, max parallelism is set such
that it's 1-2 less than the total number of installed drives. On
my ADIC Scalar 1000 with 12 drives, for example, the max
parallelism is set to 10. Whenever I run a command covering
multiple slots, up to 10 concurrent jukebox operations may
take place to process the entered command.
Terry
> -----Original Message-----
> From: George Sinclair [SMTP:George.Sinclair AT NOAA DOT GOV]
> Sent: Friday, March 07, 2003 2:51 PM
> To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
> Subject: Re: [Networker] max parallelism for a jukebox
>
> WOW! I guess I was pretty ignorant of the power of that command! It
> worked just like you said. Here's a question, though. Let's suppose
> instead of doing this:
>
> nsrjb -IEv -S 1-6
>
> to launch an inventory in parellel across all 2 drives, I instead did
> this:
>
> nsrjb -I -f /dev/rmt1 -S 1-3 &
> nsrjb -I -f /dev/rmt2 -S 4-6 &
>
> to launch a separate nsrjb process per drive. Would this not accomplish
> the same thing in the same amount of time? I guess I thought this was
> what the max parellelism value was for. To allow you to do things like
> this. I guess it allows you to do both what you were referring to and
> what I mentioned, too?
>
> George
>
> Carston Locher wrote:
> >
> > George,
> >
> > Set the "jukebox max parallelism" to "# of drives" and then do
> > the following command:
> >
> > nsrjb -IEv
> >
> > It will use all available tape drives with one nsrjb command.
> >
> > You are correct that through the GUI you must select a device.
> > However, that has never been required from the command line.
> >
> > Thanks,
> >
> > Carston
> >
> > -----Original Message-----
> > From: George Sinclair [mailto:George.Sinclair AT noaa DOT gov]
> > Sent: Friday, March 07, 2003 1:11 PM
> > To: Legato NetWorker discussion; Carston Locher
> > Subject: Re: [Networker] max parallelism for a jukebox
> >
> > I'm a little confused, and I want to make sure I'm clear about this. You
> > said:
> >
> > "For instance if you set this value
> > > to 4 and kicked off one nsrjb operation all of the 4 tape drives
> > > would be used (if you were performing operations on 4 or more slots)."
> >
> > When I run an inventory operation, I have to indicate which device to
> > use regardless of whether I'm running in the window or the command line.
> > It shouldn't matter how many slots I'm inventorying. NetWorker will only
> > use the one device provided. It will never use the other devices unless
> > I run inventories there, too. I've never seen NetWorker use any device
> > other than the one you chose regardless of how many slots you're
> > inventorying and or what the value of t he max parallelism is. Didn't
> > you mean to say that all of the 4 tape drives would be used if you were
> > performing operations from 4 separate nsrjb processes, e.g. 4
> > simultaneous inventory operations?
> >
> > George
> >
> > Carston Locher wrote:
> > >
> > > Jim,
> > >
> > > I want to clarify things because from this statement I can
> > > see you have confusion and are being misdirected.
> > >
> > > The "jukebox max parallelism" has nothing to do with backup/recover.
> > > It only has to do with nsrjb and inventory/label. It is the
> > > number of devices nsrjb will use to perform these particular
> > > operations (backup/recover). The default setting by NetWorker
> > > is "# of drives -1 (in your case would be 3). This would allow>
> > > inventory/label to use 3 devices and one would be left in reserve
> > > for backups/restores. You can change the value to 4 so all drives
> > > would be used if you like. There is no value in changing the
> > > max parallelism to be greater than the # of drives in your jukebox.
> > >
> > > Davina is actually slightly incorrect in that it is not the #
> > > of nsrjb processes allowed. For instance if you set this value
> > > to 4 and kicked off one nsrjb operation all of the 4 tape drives
> > > would be used (if you were performing operations on 4 or more slots).
> > >
> > > Please give me the name of your consultant. You don't have to
> > > send it to the group. I just want to make sure the person gets
> > > more education before they are allowed to talk to any more
> > > customers.
> > >
> > > Thanks,
> > >
> > > Carston
> > >
> > > -----Original Message-----
> > > From: Legato NetWorker discussion
> > > [mailto:NETWORKER AT LISTMAIL.TEMPLE DOT EDU]On Behalf Of Jim Lane
> > > Sent: Friday, March 07, 2003 5:35 AM
> > > To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
> > > Subject: [Networker] max parallelism for a jukebox
> > >
> > > I've just read a couple of posts on this from yesterday and I'm
> > > confused. the replies tended to suggest that this value should be
> > > related in some way to the number of drives in the jukebox that
> > > Networker should use for backup, as against leave free for ad hoc usage,
> > > restores etc. I was told by the Legato consultant that this value is the
> > > maximum number of saveset streams that can be directed to all the drives
> > > in the jukebox at once. on that basis this number is supposed to be set
> > > at or above the sum of the "target sessions" for all the tape drives. at
> > > least that's what I was told. on that theory the Legato guy set the "max
> > > parallelism" in my 4-drive STK 9710 to 32. now I'm worried. who's right
> > > here? am I over-specifying this value?
> > >
> > > Jim Lane
> > > Sr. Technical Consultant
> > > Network Services
> > > Toronto Hydro
> > > office: (416)-542-2820
> > > cell: (416)-896-8576
> > >
> > > --
> > > 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.
> > > =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
> > >
> > > --
> > > 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.
> > > =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
>
> --
> 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.
> =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
--
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.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
|
