nv-l

Re: [nv-l] MIB expression does not work correctly

2003-01-09 20:06:50
Subject: Re: [nv-l] MIB expression does not work correctly
From: Bill Evans <wvevans AT attglobal DOT net>
To: nv-l AT lists.tivoli DOT com
Date: Thu, 09 Jan 2003 20:06:50 -0500
Comment: For small numbers or precentages [if  x < 22]  and  [if (100-x) > 88] seem to be logically equvalent.  The difference is the first one can be done directly and the second requires an entry in   MIBEXPR.CONF to calculate the difference.  X in this case is the OID.  

Also, with a mib _expression_ and a comparison of  [if (x * -1) > -22] you get the equivalent and you're not restricted to numbers less than 100.  

It all depends on whom you get on the phone when you call; the math major or the one who learned about the MLM capability.  We used to teach both alternatives to the absence of less-than in the SNMP data collection thresholds.      

Todd H. wrote:
Matt.Brozowski AT tavve DOT com writes:
  
Thorsten


It looks like snmpCollect is treating your _expression_ as a
counter...
    
In my experience, every OID referenced in a MibExpr.conf file is
treated as a counter and replaced with the time rate of change 

(v(t2)- v(t1))/(t2-t1)

As a matter of practice, if an OID is not already a counter, I coerce
it to one in mib.coerce.  I'm not sure if I need to do this, but there
is some reason I learned to start doing it..... 

  
mib.coerce file.  If that doesn't work than it seems that a PMR is
your only resort... There are really two here..

1.  Why can't Thresholds do less than
    
Because MLM can, and no if there is an enhancement request for
snmpcollect to handle this, it hasn't received enough attention to get
implemented. 

  
2.  How do I coerce the type for a mib variable within an _expression_...
    
an entry in mib.coerce  will take a given OID that is an integer and
can coerce it into a counter. 

  
It is important to note here however that problem number 2 results from a
poorly defined mib rather than an actual problem in NetView...  the dhcp
mib should never have made noAddFree a counter.. it should have been a
gauge to begin with...

I'm not sure the best way to create a PMR except to ask for a
mechanism for thresholding when a counter variable goes below a
certain value.  (This seems contradictory however because counters
aren't supposed to decrease.)
    
Pose your PMR as  "How do I do.... with NetView?"  

I did... and they will tell you to use MLM's capability.  MLM can
handle a < Is there a reason you don't want to use MLM?