[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference azur::mcc

Title:DECmcc user notes file. Does not replace IPMT.
Notice:Use IPMT for problems. Newsletter location in note 6187
Moderator:TAEC::BEROUD
Created:Mon Aug 21 1989
Last Modified:Wed Jun 04 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:6497
Total number of notes:27359

5977.0. "Traps from multi-homed systems" by NYOS01::PLUNKETT () Wed May 04 1994 12:47

    I've got a problem at a customer.  We're trying to recieve traps
    from some RS/6000's (booo, hissss), and we get them with the sink
    running, no problem, but the notification request I created for all
    snmp entities in that domain never fires.  From the FCL a getevent
    on any configuration event works fine, but no iconic map
    notification.
    
    However, these RS/6000's are multihomed.  They have an ethernet
    interface, and an X25 interface.  They've been registered in DECmcc
    with the IP address of the X25 adapter, but the trap reports in with
    the ethernet interface's IP address.  Is this my problem?
    
    -Craig
T.RTitleUserPersonal
Name		DateLines
5977.1multi-homed trap solution, next questionNYOS01::PLUNKETTWed May 04 1994 13:4814
    Well, I answered my own question, which prompted another.  I did a
    little experiment, deleting the RS/6000 in question, and
    re-registering it using the IP address of the LAN adapter.  Now the
    traps and notifications work ok.
    
    Now the question is, can I force the RS/6000 to report traps with
    the IP address of their X25 adapters?  This is a question for
    somebody familiar with the SNMP daemons on AIX.  Or does the
    standard say which ip address has to be used when reporting traps. 
    The index numbers for the interfaces are: 1 is the software loopback
    lo0, 2 is the ethernet adapter, and 3 is the x25 adapter.  Anybody
    care to take a shot?
    
    -Craig
5977.2CSOADM::ROTHWhat, me worry?Wed May 04 1994 22:5914
Same problem I ran into with Cisco boxes with multiple paths... I ended
up putting each and every possible IP address in as a separate entity...
I found no way to force the Cisco to use a single address each time.

In a network with 100 LANs and 170+ T1 lines it was unacceptable to get
an event that simply said "Link down CLEVE-HUB1 Interface 22"- I had to
have somthing understandable. I gleaned info from the trap by reading
that blasted data file (don't remember the name offhand...it is passed as
P8 I think) and getting the index number- that was the only way I could
figure it out. Then, I did a lookup in an indexed file and came up with
an 'English language' name for the link that went down and then sent it
(with a few other tidbits) to the event collector.

Lee
5977.3Good Idea, I need that, tooMSDOA::REEDJohn Reed @CBO, DTN:367-6463, KB4FFE, SouthEastThu May 05 1994 00:314
    Great,  could you post your command file that does that ??
    
    
    JR
5977.4RS/6000 trap problem resolutionNYOS01::PLUNKETTTue May 10 1994 18:199
    Well the resolution on the RS/6000 issue was to de and re-register
    the entities with their new IP addresses, so that everybody on the
    MCC system, (UCX hostname, Sender Addresss on SNMP trap, Address
    component of MCC entity) agreed, and all kinds of cool, customer
    pleasing beeps and flashes started going off.  This was the extent
    of my research into the issue, but I guess the general principle
    that consistency among databases is goodness is illustrated here.
    
    -Craig