InterMapper-Talk Digest - Wednesday, April 18, 2001
Re:Annunciator Interface Request
by "Richard E. Brown" <[EMAIL PROTECTED]>
IM 3.1a10: mach-o?
by "Chris Pepper" <[EMAIL PROTECTED]>
Re: Strange periodic pattern in up times for Nortel BPS 2000
by "Doug Weathers" <[EMAIL PROTECTED]>
----------------------------------------------------------------------
Subject: Re:Annunciator Interface Request
From: "Richard E. Brown" <[EMAIL PROTECTED]>
Date: 17 Apr 2001 10:00:52 EDT
--- Paul Piecuch wrote:
.... but how about an interface/annunciator enhancement?
Our devices have names that identify TYPE-LOCATION-INDEX, as determined by
our OSI layer 1&2 staff. ( example: R-H0-C1 = Router, HT building - basement
level, core device )
The feature request is a hook into the Speech control panel so that
individual devices can be scripted with customized vocal feedback,
This would allow my NOC staff to become more creative and less crabby.
--- end of quote ---
I'm all for making the NOC Staff less crabby ...
This ought to be AppleScriptable. A notifier could call an AppleScript that
speaks the text. Does anyone on the list have a script that does this?
I'll also add this as a request for a built-in feature of InterMapper. Thanks!
Rich Brown [EMAIL PROTECTED]
Dartware, LLC http://www.dartware.com
25 S. Main St, PO Box 130 Telephone: 603-643-2268
Hanover, NH 03755-0130 USA Fax: 603-643-2289
----------------------------------------------------------------------
Subject: IM 3.1a10: mach-o?
From: "Chris Pepper" <[EMAIL PROTECTED]>
Date: Tue, 17 Apr 2001 17:32:55 -0400
ReleaseNotes.html sez:
>InterMappertm 3.1a7 Release Notes - 11 April 2001
> [FEATURE] InterMapper for MacOS X is now compiled as a Mach-O
>executable.
You mean 'can be' compiled as Mach-O, right? I was looking
for a newer mach-o than a6, but this one looks like CFM to me.
Chris
--
Chris Pepper: <http://www.reppep.com/~pepper/>
Rockefeller U Computing Services: <http://www.rockefeller.edu/>
Mac OS X Software: <http://www.mosxsw.com/>
----------------------------------------------------------------------
Subject: Re: Strange periodic pattern in up times for Nortel BPS 2000
From: "Doug Weathers" <[EMAIL PROTECTED]>
Date: Tue, 17 Apr 2001 16:16:02 -0700
The puzzle is pretty much solved.
It turns out that the IP address of the BPS 2000 switch that's beaconing was once
assigned to a different BPS 2000 switch. That switch is now part of a switch stack.
When you stack these switches, their individual IP address are supposed to
"disappear", to be replaced by an address that applies to the entire stack as a whole.
However, this doesn't seem to work correctly in the current version of the firmware -
the stack will ARP for that IP address, but won't respond to IP requests at the
address it just ARPed.
I tracked it down with a Linux program called arping. I was seeing two hardware
addresses responding to an ARP query for the IP address. One of them was correct, and
the other one was coming from a different piece of Bay equipment.
Using SNMP Watcher (Bridge MIB/dot1dTpFdbTable), I walked through my switches until I
found where that MAC address was coming from, then went there with my laptop, plugged
into the serial ports of each switch in turn, and found that one of them had been set
to the same address as the problem device. I set the switch address to 0.0.0.0 -
however, the problem is still there.
I think I'll have to bounce the offending switch stack, which I can't do without lots
of advance warning. Probably I'll just change the IP address of the beaconing switch.
Nobody cares about the switch address but me :)
Oh yeah - the 17-minute periodicy I assume has something to do wih the expiration time
of the arp cache in the InterMapper Mac or in an intermediate routing switch. When
the ARP cache expires, the next time that IP address is polled it's potluck which
device will respond to the ARP request. Get the wrong one, and the device will be
down until the next time the ARP cache expires, when you get another chance at getting
the correct MAC address.
Lessons learned (feature requests):
I had a terrible time finding Mac-based ARP tools. OTTool from Neon Software has an
ARP cache viewer, but it's unstable and tends to kill my Mac. There's something
called Network Toolbox, but it's pretty crude. iNetTools, from WildPackets, has an
ARP cache utility - in the Windows version of iNetTools (aargh). WhatRoute doesn't
have it, neither does MacPing. Anyway, I found arping and asked a Linux-compatible
coworker to install it for me. It was perfect. A Mac version of arping would be very
useful indeed.
It's hard to find MAC or IP addresses in SNMP Watcher windows. It would be nice to be
able to sort those windows by clicking on the column titles, or to have a Find
command.
InterMapper's various notification features allowed me to notice the 17-minute
periodicity. Wonderful software, InterMapper. The only feature request I can make
here is a way to find a MAC address on a map like the way you can currently find an IP
address. This find feature would be able to hunt through the dot1dTpFdbTable on each
switch, looking for matches. It could simply highlight each switch that knows about
that MAC address, letting the user figure out the network path, or it could actually
figure out the path for you and show the switch and port where that device must live.
It occurs to me that this sort of algorithm could also be used to automatically
construct a map of how all the switches are connected to each other - a feature I'm
sure is already on the enhancement request list. Also, InterMapper could find a given
IP device on the network, even if it's not on the map, by first ARPing for the MAC
address and then tracking it down as above.
>>> [EMAIL PROTECTED] - 4/11/01 9:53 AM >>>
Folks,
Got a strange puzzle I thought you'd like to see.
I have a bothersome Nortel Business Policy Switch 2000 on my network that keeps
changing between green and flashing red on my Intermapper maps. It continues to work
perfectly as a switch - my users never complain about slowness or disconnections.
When it's flashing red, I can't even ping it. I've tried pinging it from several
machines - it's as if it's not even there.
Eventually it comes back up as if nothing was wrong.
It's serving as the link between the new gigabit backbone network and the older 100mb
backbone network. It has one gigabit port leading to a Passport 8600 switch (the new
network) and a 100mb port leading to an Accelar 1200 (the old network). The traffic
on the Accelar 1200 port never seems to top 30%, so I don't think it's high traffic.
Finally I decided to set a notification to email me when the device went up or down.
This morning I had a bunch of messages in my mailbox from Intermapper, and I noticed a
strange pattern.
When the device goes down, it seems to have been running a random amount of time - at
least I couldn't see a pattern. But when it comes back up, it's always been down for
17 minutes and 49 seconds, give or take a few, or some multiple of that amount.
Here are the relevant lines from the Intermapper log (after I ran it through Excel to
filter out the irrelevant lines):
10-Apr 2:01:38 DOWN BPS 2000 (24) (Was up for 2 hours, 6 minutes, 3 seconds)
10-Apr 2:37:30 UP BPS 2000 (24) (Was down for 35 minutes, 52 seconds)
10-Apr 3:13:41 DOWN BPS 2000 (24) (Was up for 36 minutes, 6 seconds)
10-Apr 3:31:33 UP BPS 2000 (24) (Was down for 17 minutes, 52 seconds)
10-Apr 6:49:59 DOWN BPS 2000 (24) (Was up for 3 hours, 18 minutes, 26 seconds)
10-Apr 7:07:50 UP BPS 2000 (24) (Was down for 17 minutes, 51 seconds)
10-Apr 9:14:23 DOWN BPS 2000 (24) (Was up for 2 hours, 6 minutes, 29 seconds)
10-Apr 9:50:09 UP BPS 2000 (24) (Was down for 35 minutes, 46 seconds)
10-Apr 11:38:41 DOWN BPS 2000 (24) (Was up for 1 hour, 48 minutes, 32
seconds)
10-Apr 11:56:30 UP BPS 2000 (24) (Was down for 17 minutes, 49 seconds)
10-Apr 14:20:41 DOWN BPS 2000 (24) (Was up for 2 hours, 24 minutes, 11
seconds)
10-Apr 14:38:30 UP BPS 2000 (24) (Was down for 17 minutes, 49 seconds)
10-Apr 15:14:45 DOWN BPS 2000 (24) (Was up for 36 minutes, 15 seconds)
10-Apr 15:32:34 UP BPS 2000 (24) (Was down for 17 minutes, 49 seconds)
10-Apr 15:50:41 DOWN BPS 2000 (24) (Was up for 18 minutes, 7 seconds)
10-Apr 16:08:29 UP BPS 2000 (24) (Was down for 17 minutes, 48 seconds)
10-Apr 17:38:43 DOWN BPS 2000 (24) (Was up for 1 hour, 30 minutes, 4
seconds)
10-Apr 17:56:37 UP BPS 2000 (24) (Was down for 17 minutes, 54 seconds)
10-Apr 19:27:04 DOWN BPS 2000 (24) (Was up for 1 hour, 30 minutes, 27
seconds)
10-Apr 19:44:53 UP BPS 2000 (24) (Was down for 17 minutes, 49 seconds)
10-Apr 22:09:23 DOWN BPS 2000 (24) (Was up for 2 hours, 24 minutes, 30
seconds)
10-Apr 22:26:52 UP BPS 2000 (24) (Was down for 17 minutes, 24 seconds)
10-Apr 23:03:18 DOWN BPS 2000 (24) (Was up for 36 minutes, 7 seconds)
10-Apr 23:57:01 UP BPS 2000 (24) (Was down for 53 minutes, 43 seconds)
I have other BPS 2000s on the network with the same firmware rev and they don't do
anything like this - they just stay up all the time like good little switches.
Does anyone have ANY idea what might be going on here?
This may or may not be relevant, but 1024 seconds is 17 minutes 4 seconds. With a
poll window of 30 seconds, could that look like 17 minutes 49 seconds?
Have fun, and thanks in advance for any clues you can give me.
Doug
____________________________________________________________________
Note: To unsubscribe from this mailing list, please send email to:
<mailto:[EMAIL PROTECTED]> Thanks!
----------------------------------------------------------------------
End of InterMapper-Talk Digest
____________________________________________________________________
Note: To unsubscribe from this mailing list, please send email to:
<mailto:[EMAIL PROTECTED]> Thanks!
