[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

2834.0. "Notification - Potentially Serious Problem re: adjacency up/down" by PJWL::LAMB (Peter Lamb - GSG Santa Clara MAIL=MUTTON::LAMB) Thu Apr 23 1992 11:26

Hello,

I've been working for several weeks now trying to do 
something which obstensibly should be fairly easy now
with the advanced notification capabilitites of 1.2.

Picture the following configuration...

     ---------       ------------         -------------
     | DECmcc|       | DECnet   |         |  DECnet   |
     | v1.2.7|       | IV Router|         | End NOde  |
     |       <----------sink    |         |           |
     ---------       ------------         -------------
          |                |                   |
 -------------------------------------------------------

All three nodes are on the same lan.  The DECmcc system
is successfully receving all class 4 events from the 
router.  When I take the DECnet End Node up and down I
am able to see 4.14 "Node Reachability Change" events and
through the use of targeting even toggle the correct node.
However, while I do receive adjacency messages for a few
other nodes on the network I NEVER receive them for my
adjacent node.  

I'm now suspecting that the problem is with DECnetVAX and
that if this is true, it (in my opionion) significantly
reduces the usefullness of DECmcc event notification.

Something I read in the DECnet VAX notes files led me to
believe that some sort of agreement was reached not to
report adjacency messages on nodes on the same physical
ethernet as this would cause too many messages/traffic.

I have a note in the DECnet VAX conference to attempt to
confirm this but as of yet have not gotten any replies 
(see #4332)

Anyone else care comment??!!

Thanks!

Peter

T.R	Title	User	Personal Name	Date	Lines
2834.1	make MCC a router???	ICS::WOODCOCK		`Thu Apr 23 1992 11:43`	18
	Hi Peter, > Something I read in the DECnet VAX notes files led me to > believe that some sort of agreement was reached not to > report adjacency messages on nodes on the same physical > ethernet as this would cause too many messages/traffic. Interesting, I hadn't heard about this one before. One thing you could try is to make the MCC node a router (it's an end node now right??). With V1.1 we were using adjacency events to keep track of both circuit and node outages. Our MCC system was a level 1 router on a large LAN. By default with MCC you set your own node to pass the adjacency events from itself to MCC even when you primarily want other routers to send you the status events. When this was done in PKO the adjacency events from the PKO LAN had us jumpin' so I suspect it will also work for you. best regards, brad...
2834.2	Ultrix=no routing	PJWL::LAMB	Peter Lamb - GSG Santa Clara MAIL=MUTTON::LAMB	`Thu Apr 23 1992 12:34`	13
	Hi, Thanks for the suggestion! However, I forgot to mention one thing... My DECmcc system is Ultrix based ie. no router capability. Note: this still doesn't explain why my DECNET IV router does not see adjacency messages for my system even throug opcom messages (note I know I'm not blocking these messages since I do see adjacency messages for other systems on the net.) Peter
2834.3	Router Router where's my router	MUTTON::LAMB	Peter Lamb - GSG Santa Clara	`Thu Apr 23 1992 14:13`	32
	Well, I have looked into this a little further and apparently the reason I'm not getting adjacency messages is because my router is not showing up as the adjacent router. On the router I am sinked to SHOW NODE MUTTON (the node I wish to receive adjacency requests from reveals that SCRTR2 not PJVAX is the next hop to node mutton even though they are sitting on the same thin wire segment less then 4 feet away from one another!! Node State Active Delay Circuit Next node Links 30.352 (MUTTON) reachable 1 1 QNA-0 30.855 (SCRTR2) So I guess if I had the priviliges to sink to this router everything would work. However, we did some further testing and by shutting down one router the adjacency moves to another router (providing one is available) and never comes back... For example, say we have end node A and two routers B & C (all three on the same cable). If A is currently adjacent to both B & C shutting down B shows that A is now adjacent to C, so far so good. However, restarting B does not re-establish adjacency with A such that C is now the only node that is adjacent to A. Thus, it appears that to consistantly receive adjacency messages one must be sinked to every router on the ethernet network! Peter
2834.4	Problem Solved (sort of)	MUTTON::LAMB	Peter Lamb - GSG Santa Clara	`Thu Apr 23 1992 19:53`	15
	Well i figured out how to get adjacency messages... I needed to increase the number for NCP EXE MAX BROADCAST NONROUTERS Upon doing this the nodes on my cable suddenly became adjacent and I started receving adjacency up/down messages. Now if I can only keep notification running.... Regards, Peter Lamb
2834.5	Set an alarm on it :-)	TOOK::R_SPENCE	Nets don't fail me now...	`Thu Jun 04 1992 14:52`	6
	That parameter is a great one, by the way, to set a wildcard alarm rule on. Have it run once a day at 2am or something and you will discover if people still have the out of box parameter set. s/rob
2834.6	over 100,000 events and still running....	TOOK::CALLANDER	MCC = My Constant Companion	`Thu Jun 18 1992 12:38`	5
	Our system testing on the pre SSB kit has over 100,000 events and it keeps on ticking (note it is also show NO memory leaks). jill (thought you might be interested....)