[mrtg] Re: -MS SNMP stuff
charles at toucan.ie
Mon Aug 9 19:24:38 MEST 1999
Have finally got round to looking at part of this. The dual processor
machine I wanted to try Win2K server on has been appropriated for other
purposes so I'll have to wait a while to try that. I did try Win2K
workstation beta3 on a single processor machine and the snmp service does
support the host resources mib without needing any extra configuration. So
that looks to be a good sign anyway. There's also a whole new set of MS
enterprise mib variables that I have yet to find a mib file for but I'm
going to keep looking...
From: Neil Francis [mailto:n.j.francis at bath.ac.uk]
Sent: 29 July 1999 15:34
To: Charles Gillanders
Subject: Re: RE: -MS SNMP stuff
No - I have only looked at NT 5 workstation. Not Server and not SNMP.
It would be interesting to see how you fare.
On Mon, 26 Jul 1999 14:49:11 +0100 Charles Gillanders
<charles at toucan.ie> wrote:
> Thanks for the info, you may be interested to know that I saw a posting to
> the effect that NT5 will natively support the host-resources mib properly
> without needing to add performance manager counters into the equation.
> I am going to check with beta 3 later this week (time allowing) I can let
> you know what I find, unless you've already done this.
> -----Original Message-----
> From: Neil Francis [mailto:n.j.francis at bath.ac.uk]
> Sent: 26 July 1999 13:11
> To: Charles Gillanders
> Subject: Re: -MS SNMP stuff
> I had a couple of responses. I think the root cause is Microsoft's
> half hearted implementation of SNMP on its NT OS. The second response
> here is probably the root cause.
> Also I have _always_ had a problem with the % counters on NT boxes.
> They either show 100% or nothing. Even for disk usage, memory useage
> and CPU. Again - I think it is just bad coding from MS. The only way
> to so this is manually.
> Hello Neil!
> I was hunting for information on the exact same subject, and came across
> your post dated Jan 28th.
> According to MS, the SNMP counters for SMP machines will not be
> reliable, period. Something to do with MS deciding not to implement
> proper locking when fetching the data. Check out TechNet, searching for
> "SMP and SNMP".
> I would *love* to be mistaken, because I have the same problem to solve.
> If you can help out, please don't hesitate!
> Fernando da Silveira Montenegro Nutec Servicos Corporativos
> System/Network Consultant Sao Paulo, SP, Brazil
> mailto:montenegro at nutec.com.br http://www.nutec.com.br
> voice.:+55-11-5505-5728 #include <std_disclaimer.h>
> I saw your posting as i was searching for
> the key works SNMP 100% looking for others
> who have seen what i have seen.
> Yes it is true on a dual system i am
> finding that usualy processor 2 spikes to 100
> while using snmputil.exe Here is why its doing
> it. Inside of snmputil it calls SnmpMgrOpen()
> to open a connection to the snmp server this
> seems to be the problem call. If i put a Sleep(300)
> after this call i get more realistic percentages
> from my system. So the bottom line is that
> the SNMP.exe serverice on the NT Server I am
> monitoring was written by someone who does
> not know how to write efficient C code. I plan
> on writing to Microsoft on this one but I doubt
> they will do anything.
> I also tried sp4 so that didnt help.
> If you have anymore info on this please let
> me know..
> Sean Mathews
> On Mon, 26 Jul 1999 11:46:45 +0100 Charles Gillanders
> <charles at toucan.ie> wrote:
> > Some time ago you wrote this message in comp.protocols.snmp I was
> > wondering if you ever got anything sorted out, I am seeing similar
> > issues here both on dual and quad boxes, it doesn't seem to matter
> if > they're PII or PIII either.
> > > I would very much appreciate anything you discovered about this.
> Oh > and also have you ever discovered any way of averaging the
> processor > figures to show average load, I tried Process \ Total \
> %Processor > Time but that always shows up 100%.
> > > Thanks,
> > > Charles
> > >
> Neil Francis n.j.francis at bath.ac.uk
> Tel: (44) 122 532 3571 Bath University Computing Services
n.j.francis at bath.ac.uk
Tel: (44) 122 532 3571
Bath University Computing Services
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the mrtg