What do you class as a thread - you could think of each "pinger" as a
separate thread in my opinion. The -q options allows you to set the number
of concurrent pingers which netmon can use. This is evident in a netmon
trace. From memory the default is 16 pingers and if you have "sufficient"
hardware it would not be unreasonable to increase this to 32 to see what
affect it has. You are probably not going to get any real benefit from
going beyond 32 pingers - I believe in the past it has been a
recommendation from the NetView Performance group to not exceed 32 pingers
but I maybe wrong.
Increasing the number pingers should definitely be done after ensuring that
adjustments to your status polling interval, number of retries, and timeout
values have all been tuned and retuned and retuned ..... where possible.
Also ensure your DNS Server is working efficiently (include RES_TIMEOUT=1
and RES_RETRY=1 in your environment file to reduce delays caused by poor
DNS response).
The last thing I would look at if you are in a Unix environment is your
operating system's ARP settings. Make sure your ARP queue size and other
ARP settings are adequate for your managed environment.
The -Q option for SNMP polling is new in version 6 and will be similar in
concept to the -q option but don't go using the values I talk about above -
seek advise from support about tuning this one (or use your test
environment if you have one). Both -q and -Q are available in NetView for
Unix to answer a previous question.
There was also talk of a "local Linux poller at each data center" - sounds
like functionality that MLMs already provide. The MLM code is provided
with the NetView media and included in the licence as far as I know.
And finally, the upgrade to 6.0.1 was from what version ? Note you maybe
moving from a version of NetView that use to auto adjust the timeout values
for status polling, usually down to 1 second for nodes that consistently
responded and I don't think 6.0.1 does this anymore - what you set is what
you get - this may be adding time to your status polling cycle - a long
shot but doesn't hurt to check it all. Are you seeing lots of status
changes on your maps ? Do you have lots of nodes in your network going up
and down ? What kind of response time do you see for pings from your
NetView box to remote nodes ? Study your netmon trace as it can tell you a
lot. I hope something here helps.
Gareth Holl
Software Engineer
gholl@tivoli.com
Tivoli Systems / IBM Corporation
Research Triangle Park, North Carolina. 1-800-TIVOLI-8
"Peter Anderson" <pETERanderson@westpac.com.au>@tkg.com on 01/10/2001
06:39:57 PM
Please respond to IBM NetView Discussion <nv-l@tkg.com>
Sent by: owner-nv-l@tkg.com
To: "Netview List (E-mail)" <nv-l@tkg.com>
cc:
Subject: RE: [NV-L] netmon question
It is queue size not number of threads on Unix.
>From netmon man page:
-q number
Sets the queue size for ICMP pings. The value you specify with this
switch can degrade system performance and response time; therefore,
using Server Setup to regulate netmon behaviour is recommended.
regards,
Peter Anderson
Senior Communications Analyst
Westpac Banking Corporation
Any views or opinions presented are solely those of the author and do
not necessarily represent those of Westpac Banking Corporation.
-----Original Message-----
From: Boyles, Gary P
Sent: Wednesday, 10 January 2001 23:24
To: 'IBM NetView Discussion'
Cc:
Subject: RE: [NV-L] netmon question
Sarah,
Maybe I mis-spoke (of you're on UNIX).
On NetView NT there is the -q options (to specify the number of
ping-threads),
and -Q to specify the number of SNMP threads.
Sarah,
Maybe I spoke to soon...
I don't know if netmon supports this on UNIX. It is on NetView NT.
Do a netmon -h (or netmon -?) and see if it is listed.
It was on NV NT.
If so... modify your /usr/OV/lrf/netmon.lrf file and restart netmon.
Maybe someone from 3rd-level support can chime in here.
Regards,
Gary Boyles
-----Original Message-----
From: Sarah Romeis [mailto:sromeis@us.ibm.com]
Sent: Wednesday, January 10, 2001 2:37 PM
To: IBM NetView Discussion
Subject: RE: [NV-L] netmon question
How do you increase the number of ping threads with Netview?
Sarah Romeis
IBM Global Services 1630 Long Pond Rd Rochester, NY 14626
Internet: sromeis@us.ibm.com
Lotus Notes: IBMUSM10(SROMEIS)
Voice: (716) 723-4354 Fax:(716) 723-4299
"Boyles, Gary P" <gary.p.boyles@intel.com>@tkg.com on 01/10/2001 04:31:27
PM
Please respond to IBM NetView Discussion <nv-l@tkg.com>
Sent by: owner-nv-l@tkg.com
To: "'IBM NetView Discussion'" <nv-l@tkg.com>
cc:
Subject: RE: [NV-L] netmon question
Have you tried increasing the number of ping-threads?
Gary Boyles, Intel
-----Original Message-----
From: Stephen Elliott [mailto:selliott@epicrealm.com]
Sent: Wednesday, January 10, 2001 1:19 PM
To: 'nv-l@tkg.com'
Subject: [NV-L] netmon question
Hello, everyone,
Here's the teaser question of the week: Has anyone tried to, or know of
someone who has tried to, modify netmon to use fping vs. ping? I'm dealing
with consistent netmon backlogs since the v6.0.1 upgrade (it used to stay
current. We have a PMR open on this issue.) and we can't reduce the status
interval. We're looking at several options (one being a local Linux poller
at each data center, etc.) to get around it but was just wondering if this
had been tried.
Regards,
Steve Elliott
Sr. Network Mgmt. Engineer
epicRealm, Inc.
214-570-4560
_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l
_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l
_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l
_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l
_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l
|