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

Conference netcad::hub_mgnt

Title:	DEChub/HUBwatch/PROBEwatch CONFERENCE
Notice:	Firmware -2, Doc -3, Power -4, HW kits -5, firm load -6&7
Moderator:	NETCAD::COLELLADT

Created:	Wed Nov 13 1991
Last Modified:	Fri Jun 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	4455
Total number of notes:	16761

3777.0. "DE500 and DEC Repeater 900Tm hight collision rate" by CX3PST::NOTAMI::A_ANDERSON (CX03 2/H13 NSU/VAX MacGhille Aindrais) Tue Aug 06 1996 20:06

Hello all,

I have a customer that just purchased 100 Ventrus 575 PCs with DE500 100/10
ethernet interfaces.  He has all of these PC in a remote building connected to
three DECrepeater 900TM's in a DEChub900. The DEChub900 has a DR90FA with a
fiber link to another building DEChub900 and a DECrepeater 90FL.

After 2-3 weeks he noticed a high rate of collisions on the LAN.  High enough to
show up as send failures due to excessive collisions.  

If he disconects the Fiber link the errors go away.  If he removes all three
DR900TM the errors go away.  If he stops using the New PC's the errors go away. 
As soon as he starts using the network from one of the PC's the Collision rate
starts increasing.  He normally expects to see 1 collision in 100 packets.  He
is using a Network General Sniffer to measure this.

Are there any known issues with DE500's and DEC repeater 900TM's?

I was told to verify that the DE500 is in half duplex mode.  Any ideas on how
you would do this on a PC?

The DR90FA has the phase lock loop, (Feature) and is sensitive to some Nic
cards.  Is the DE500 one of those cards?

I am open to any and all suggeastions
Alan S. Anderson
Network Support CSC CS

T.R Title User Personal
Name Date Lines

3777.1 A few observations NETCAD::GILLIS Tue Aug 27 1996 18:09 29

T.R	Title	User	Personal Name	Date	Lines
3777.1	A few observations	NETCAD::GILLIS		`Tue Aug 27 1996 18:09`	29
	Some observations: 1) By using all repeaters in your configuration, your customer has created a single "collision domain". It sounds like you are either at the limit of the maximum repeater hop count (4) or beyond it for stations connected at opposite ends of your extended collision domain. If this editor provided a graphical tool, I'd draw what I thought your network looked like. 2) Are there a lot of runt frames and alignment errors seen on the sniffer as well? These are also indicative of excessive collisions 3) Are there any jabbers captured by the sniffer, or listed by any of the repeaters anywhere in the path of your topology? It's indicative of faulty repeater hardware and sometimes faulty wiring. 4) Is there a loop in the network? 5) See if you can place a bridge between the two hubs. This will do a number of things for you: a) reset the repeater hop count to zero once traffic leaves the bridge to the next hub. This creates separate collision domains local to each of the two hubs. b) if you have a loop, will block other interface Good luck. If all else fails, send me email directly, and we'll talk John Gillis clearVISN developer

Some observations:

1) By using all repeaters in your configuration, your customer has created
a single "collision domain". It sounds like you are either at the limit
of the maximum repeater hop count (4) or beyond it for stations connected
at opposite ends of your extended collision domain. If this editor provided
a graphical tool, I'd draw what I thought your network looked like.

2) Are there a lot of runt frames and alignment errors seen on the sniffer
as well? These are also indicative of excessive collisions

3) Are there any jabbers captured by the sniffer, or listed by any of
the repeaters anywhere in the path of your topology? It's indicative
of faulty repeater hardware and sometimes faulty wiring.

4) Is there a loop in the network?

5) See if you can place a bridge between the two hubs. This will do a number
of things for you:
	a) reset the repeater hop count to zero once traffic leaves
	   the bridge to the next hub. This creates separate
	   collision domains local to each of the two hubs.
	b) if you have a loop, will block other interface

Good luck. If all else fails, send me email directly, and we'll talk

John Gillis
clearVISN developer