Veritas-bu

[Veritas-bu] Is there an official document of client and serv er N ICand switch settings

2004-03-18 15:56:31
Subject: [Veritas-bu] Is there an official document of client and serv er N ICand switch settings
From: Dwayne.Brzozowski AT mail.va DOT gov (Brzozowski, Dwayne)
Date: Thu, 18 Mar 2004 14:56:31 -0600
If your CE card is fiber, then what I have described below will happen. It
has happened on 3 different servers now that have more than one CE card. I
didn't understand this at first a few months ago when we got our first CE
card with a new server,  and I helped crash this server in the middle of the
day because the /  filesystem was filled with link messages that spewed all
weekend long when I made, what I thought would work, a global CE interface
change, like I would with a hme card. Big mistake thinking logically. If you
have a CE 10/100, everything should be ok. As  you can see from the file
below, every ce card on this system is in the tuning file, no matter if it
is plumbed or not. 

-----Original Message-----
From: Paul Boatman [mailto:boatm005 AT groupwise.umn DOT edu]
Sent: Thursday, March 18, 2004 2:37 PM
To: Brzozowski, Dwayne; veritas-bu AT mailman.eng.auburn DOT edu
Subject: RE: [Veritas-bu] Is there an official document of client and
server N ICand switch settings


Dwayne,

We are not currently running GigE yet... am I misunderstanding you ?

Thanks,

Paul

>>> "Brzozowski, Dwayne" <Dwayne.Brzozowski AT mail.va DOT gov> 3/18/2004
14:35:05 >>>
Paul,
The fiber version (gig) of this interface has some quirky behavior. 
Even
when the interface has not been plumbed, the interface will spew link
up/down messages continuously to the console if one of two conditions
are
not met.  1.) a cable must be attached to the interface --or -- 2.)
the
interface must be forced to auto-negotiate on (yes, that is on).  This
condition will not occur if the interface is being used normally.  That
is,
plumbed, cable inserted and auto-negotiate=false

I have my CE card settings in /etc/rc2.d. I have mine put into a file
called: 
S68net.tune

contents are:
#!/sbin/sh
ndd -set /dev/ce instance 0
ndd -set /dev/ce adv_autoneg_cap 0
ndd -set /dev/ce adv_1000fdx_cap 1
ndd -set /dev/ce adv_1000hdx_cap 0

ndd -set /dev/ce instance 1
ndd -set /dev/ce adv_autoneg_cap 0
ndd -set /dev/ce adv_1000fdx_cap 1
ndd -set /dev/ce adv_1000hdx_cap 0

ndd -set /dev/ce instance 2
ndd -set /dev/ce adv_autoneg_cap 1
ndd -set /dev/ce adv_1000fdx_cap 1
ndd -set /dev/ce adv_1000hdx_cap 0

ndd -set /dev/ce instance 3
ndd -set /dev/ce adv_autoneg_cap 1
ndd -set /dev/ce adv_1000fdx_cap 1
ndd -set /dev/ce adv_1000hdx_cap 0

ndd -set /dev/ce instance 4
ndd -set /dev/ce adv_autoneg_cap 1
ndd -set /dev/ce adv_1000fdx_cap 1
ndd -set /dev/ce adv_1000hdx_cap 0

ndd -set /dev/ce instance 5
ndd -set /dev/ce adv_autoneg_cap 1
ndd -set /dev/ce adv_1000fdx_cap 1
ndd -set /dev/ce adv_1000hdx_cap 0
ndd -set /dev/ce instance 6
ndd -set /dev/ce adv_autoneg_cap 1
ndd -set /dev/ce adv_1000fdx_cap 1
ndd -set /dev/ce adv_1000hdx_cap 0

ndd -set /dev/ce instance 7
ndd -set /dev/ce adv_autoneg_cap 1
ndd -set /dev/ce adv_1000fdx_cap 1
ndd -set /dev/ce adv_1000hdx_cap 0

ndd -set /dev/ce instance 8
ndd -set /dev/ce adv_autoneg_cap 1
ndd -set /dev/ce adv_1000fdx_cap 1
ndd -set /dev/ce adv_1000hdx_cap 0

ndd -set /dev/ce instance 9
ndd -set /dev/ce adv_autoneg_cap 1
ndd -set /dev/ce adv_1000fdx_cap 1
ndd -set /dev/ce adv_1000hdx_cap 0

-djb

-----Original Message-----
From: Paul Boatman [mailto:boatm005 AT groupwise.umn DOT edu] 
Sent: Thursday, March 18, 2004 7:33 AM
To: Brzozowski, Dwayne; veritas-bu AT mailman.eng.auburn DOT edu 
Subject: Re: [Veritas-bu] Is there an official document of client and
server N ICand switch settings


I work at the 6th largest university in the country.  About 1 year
ago,
the entire university was required to switch to AUTO-AUTO across ALL
platforms because of  network infrastructure changes ("eek !").  The
impact on Veritas Netbackup has been ridiculous.  Our drives are
sitting
here getting 10%-20% thruput, and our capacity is maxed out when it
shouldn't be.  Tape management and load balancing between 2 robots (1
of
which is maxed out tape-wise, the other is maxed out drive-wise)  is a
nightmare.  Upper management is blaming the software, scheduler, and
drive technology claiming that they are not sufficient.

Is anyone else in a similar situation ?

What are the "settings that work" Dwayne ?  I am very interested. 
Maybe we could present this info to our Network / Telecom guys.  By
the
way Dwayne, I am a veteran so I sure do appreciate you guys in the VA.

Thanks,

Paul

>>> "Biller, Tim" <Tim.Biller AT uk.experian DOT com> 3/18/2004 02:55:30 >>>
Hi Cheryl,

Have the Network Admins lock the switch port speed to 100Mbit and Full
Duplex and make sure the client NIC is set exactly the same on ALL the
servers.  The Unix boxes will need setting explicitly as Mark says,
and
a
file added to ensure they stay that way after a reboot.  (Your
sysadmins
should know this....)

Most network switches always auto-negotiate to the lowest possible
speed so
it's essential to lock this down.  Any decent network admin should
know
this....

Are you running on a dedicated backup LAN? If there's no impact on
other
systems there's absolutely no reason not to max out the ports and
NICs.
 

I'm not aware of any specific documentation that states that 100/Full
should
be used but if performance sucks under autoneg and it flies under
100/Full
then there's your argument.  What tape drives are you using?  Are you
using
SSO?

Cheers - Tim


Date: Wed, 17 Mar 2004 12:03:10 -0700
From: "King, Cheryl" <cheryl.king AT intrado DOT com>
To: "Veritas-Bu \(E-mail\)" <veritas-bu AT mailman.eng.auburn DOT edu>
Subject: [Veritas-bu] Is there an official document of client and
server NIC
and switch settings

Over and over again we have problems with poor performance with new
client
backups and restores.  Is there a document that says what the NIC
settings
should be for each client type?  For example we have highest speed
possible
for all of them, half duplex for HP, and full duplex for everything
else, I
think (I haven't checked every system and wasn't here when some were
implemented).  Setting Auto detect on the switch seems to not work
correctly
(causes slow performance).   I think the switches are set to match the
server, but have no way to verify that.  I'd like to be able to
present
an
official document to the System Administrators and the Network
Engineers
that states the correct configuration requirements.  Does Veritas have
such
a document or statement?

Is this a problem in other environments?  Is this just an issue in
Veritas
Backup environments?  Are my SAs and Network Engineers not following
some
industry, standard, best-practices?    I've been in IT for many years
but
working with Veritas NBU for 1 year.  I don't recall ever having
network
issues like this before.

NBU 4.5 FP5 Solaris 8 Master/Media
Clients OSF1_V5, HP-UX 11.00, HP-UX10.20, Windows2000, WindowsNT,
RedHat2.4,
RedHat 2.2, Solaris8, Solaris7.

Thanks in advance.

Cheryl King


From: "Donaldson, Mark" <Mark.Donaldson AT experianems DOT com>
To: "'King, Cheryl'" <cheryl.king AT intrado DOT com>,
   "Veritas-Bu (E-mail)"
         <veritas-bu AT mailman.eng.auburn DOT edu>
Subject: RE: [Veritas-bu] Is there an official document of client and
serv
        er NIC and switch settings
Date: Wed, 17 Mar 2004 12:31:48 -0700

The network settings are outside Netbackup control.  

You should always set to your highest possible speeds.  For many
server/switch combinations, auto-negotiation fails to negotiate the
highest
settings and they must often be forced to the best setting using ndd
or
similar utilities.

Half-duplex performance stinks - are you sure that's what only what HP
can
do?

Your network person should be able to query a smart-enough switch for
the
switch port settings, the SA should be able to query the ethernet port
for
the same settings.  


This e-mail has come from Experian International: winner of the UK's
National Business of the Year Award 2003.

==========================================================================
Information in this e-mail and any attachments are confidential, and
may 
not be copied or used by anyone other than the addressee, nor
disclosed

to any third party without our permission. There is no intention to 
create any legally binding contract or other binding commitment
through

the use of this electronic communication unless it is issued in
accordance 
with the Experian Limited standard terms and conditions of purchase or

other express written agreement between Experian Limited and the
recipient 
Experian Limited (registration number 653331) Registered office: 
Talbot House, Talbot Street, Nottingham NG80 1TH

Although Experian has taken reasonable steps to ensure that this
communication 
and any attachments are free from computer virus, you are advised to
take 
your own steps to ensure that they are actually virus free.

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


This e-mail and its attachments have been scanned for viruses.
EAS - Systems Support, OIT, University of Minnesota


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