["Wozny, Scott" <swozny@xxxxxxxxxxxxx>]

Thanks.  This makes much more sense.  I'll look at fixing the things
about the MIBs in the other application the NET-SNMP doesn't like and
hopefully that will correct the problem.

Thanks for all your help,

Scott A. Wozny
Enterasys, NYC

-----Original Message-----
From: Martin Regner [mailto:martin.regner@xxxxxxxxx] 
Sent: Tuesday, April 08, 2003 12:09 PM
To: Wozny, Scott; ethereal-users@xxxxxxxxxxxx
Subject: Re: [Ethereal-users] Problem with new installation.

Scott Wozny wrote:
<It's looking like you're right.  The directory that tethereal was
<reporting all the errors from is the same one that I have a MIBDIRS
<value pointing to.  I took out the MIBS and and MIBDIRS variables and
<both ethereal and tethereal started running fine (just taking out
<MIBDIRS cleared up the tethereal errors, but still kept the apps from
<running).

<I could remove the offending application and (presumably) ethereal will
<behave fine (but I kind of want to keep it around) or I could back rev
<to before 0.9.8, but I'd rather fix the problem and use the current
<version if I could.  I did some poking around the NET-SNMP project
<website on sourceforge but couldn't figure out which specific problem
<MIBS and necessary imports you were speaking about.  Do you have a link
<to the specific ASN.1 stanzas I need to add into these MIBs to make
<NET-SNMP happy with them?


What I remember I installed the net-snmp tools (www.net-snmp.org) and
used one of the tools included
(e.g. snmptranslate) and maybe put the warning level up in net-snmp so
that I could see exactly what mib that caused the problem and some
details about what was problem with that MIB.

The crash information I got on Windows 98 looked like the one in this
message:
http://www.ethereal.com/lists/ethereal-users/200212/msg00090.html

The crash occurred in the function find_tree_node. The backtrace looked
something like this:

FramePtr ReturnAd Param#1  Param#2  Param#3  Param#4  Function Name
0012fa10 0067358e 00000020 ffffffff 00c09590 00000000
ethereal!find_tree_node  (FPO: [EBP 0x00000001] [2,0,4])
0012fb28 00679958 00c09590 006693c2 0012fb58 00000000
ethereal!adopt_orphans  (FPO: [0,65,2])
0012fb30 006693c2 0012fb58 00000000 0012ff0f 00000019
ethereal!read_all_mibs  (FPO: [0,0,1])
0012fb48 005b50ff 00932ec3 00c08440 0012fb60 00601109 ethereal!init_mib
(FPO: [EBP 0x0012fb58] [0,1,4])
0012fb58 00601109 0012fb80 0064ad77 00000014 00131e70
ethereal!proto_register_snmp 
0012fb60 0064ad77 00000014 00131e70 0012fb80 0064f729
ethereal!register_all_protocols 
0012fb80 0065b378 00000000 00600bc0 0060120f 0012ff14
ethereal!proto_init 
0012fb94 0060d968 00000000 00600bc0 0060120f 0461e3a4 ethereal!epan_init

0012ff14 0060f0ed 00000001 00c00180 0012ffc0 0060c6e2 ethereal!main 

There was a message with more or less the same crash information in one
of the messages on
NET-SNMP Sourcefourge bug-reports:
http://sourceforge.net/tracker/?atid=112694&group_id=12694&func=browse

It was there I could see the hint that you could get that specific
problem if there was an IMPORT .... FROM missing in the MIB.
There is also a similar thing mentioned in the following discussion
(see: 'you need to import "enterprises" FROM SNMPv2-SMI')
http://net-snmp.sourceforge.net/irc/net-snmp.log.2002-9-12.html

I'm not sure if they really have solved the problem in a later version
of NET-SNMP.