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

Conference 7.286::fddi

Title:FDDI - The Next Generation
Moderator:NETCAD::STEFANI
Created:Thu Apr 27 1989
Last Modified:Thu Jun 05 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2259
Total number of notes:8590

1594.0. "Explanation for ring events?" by BUNDE::BURNS () Fri Feb 17 1995 10:46

I'm working with a customer who is suffering from serious network problems 
which are very costly in production losses.  I'd like some help from the 
readers of this conference with a few FDDI facts and in interpreting some 
of our observations of the behavior of this network.

There are 27 stations (MACs) currently attached to the ring.  Five of these 
27, plus two concentrators without separate MACs, are dual-attached in the 
dcr backbone; the rest are attached through the concentrators.

Among the dual attached stations are 2 DECbridge520s and 3 brand-X 10/100 
bridges.  The stations single-attached through the concentrators include 
mostly VAX 7000s, 6000s, and 4000s, 2 or 3 HP 9000s, a W&G analyser, and 
two brand-Y brouters.

The bulk of traffic is across the five 10/100 bridges between hosts 
connected to FDDI and hundreds of small processors and X-window terminals 
connected to Ethernet.

One set of observations which has gotten a lot of attention involves a 
certain type of ring event recorded repeatedly by the W&G analyser.  In a 
typical event of this sort, a SAS seemingly leaves the ring and then 
returns within in a few seconds.  Besides the fact that most of these 
incidents seem to be spontaneous, rather than the result from the station 
being intentionally removed from the ring in any way, most also proceed in 
what seems to me to be a chaotic way.

Specifically, the record shows in these cases that the ring cycles between 
operational and non-operational states three or more times, with many 
hundreds (e.g., 3000) claim frames being recorded by the W&G analyser and 
counted by stations on the network.  Upstream neighbor change messages are 
recorded, sometimes from stations already declared as missing by their down 
stream neighbors.  Within a few seconds, further messages announce the re-
detection of the lost neighbors and all signs indicate that the network has 
returned to its earlier state.

Ocassionally, there is a similiar incident recorded in which the ring 
cycles once and only a small number, close to 27, of claim frames are 
counted.  In these "orderly" cases, the neighbor change messages also seem 
reasonable.  In a test, the dcr backbone was interrupted and caused to 
wrap.  An "orderly" ring cycling was observed.  In another test, when an new 
SAS was inserted and powered on and, somewhat later, off, events of the 
"chaotic" type described above were produced.

I have these questions:

1. On a ring such as I've described, should the "chaotic" type of event 
occur, even in the case of a SAS actually being inserted?

2. In a healthy ring with 27 stations, what would be the expected range of 
numbers of claim frames counted when ring re-initialization is needed?  
What would be the range most frequently observed?

3. Has anybody seen this sort of behavior?  Is it pathological?

These observations possibly all amount to a red herring, and may not 
really be involved in the serious network disruptions which are causing the 
real grief, but we need to resolve all the questions this raises.

Please help if you can.


Malcolm
T.RTitleUserPersonal
Name		DateLines
1594.1Honest, I'm not making this up!BUNDE::BURNSMon Feb 20 1995 08:3641
Below is a fairly typical example of a report from the W&G analyser showing 
the seemingly spontaneous occurrence of the type of event I described in 
the base note.

Also, I forgot to mention that on several occasions, when data was 
carefully preserved and recorded, we were able to detect that the VMS NCP 
line "Ring initializations received" and "MAC frame count" counters on a 
VAX which was apparently "dropped out" in one of incidents increased by 
several billion in 30 minutes or less and overflowed the counter fields.  
As best we can tell, no other systems on the FDDI LAN were affected in this 
way at that same time.

_____________________________________________
 02:15:54	Int	Detected 1 IFG(s) less than 6 bytes in last second.   
 02:31:07	MAC	Ring NOT operational.                                 
 02:31:07	MAC	Ring operational.                                     
 02:31:07	MAC	Ring NOT operational.                                 
 02:31:07	MAC	Ring operational.                                     
 02:31:07	MAC	"Detected 839 CLAIM frame(s) in last second, OLD_TNEG:
			 5000 uS, NEW_TNEG: 5000 uS."
 02:31:08	MAC	Ring NOT operational.                                 
 02:31:08	MAC	Ring operational.                                     
 02:31:08	SRF	PORTPathChange Event at Station 00-00-1D-10-5E-73 PORT 
			Index = 5.
 02:31:09	MAC	"Detected 2139 CLAIM frame(s) in last second, 
			OLD_TNEG: 5000 uS, NEW_TNEG: 5000 uS."
 02:31:09	NIF	UNA change at station 08-00-2B-3B-FA-2D OLD_UNA: 
		        00-80-16-08-80-9E  NEW_UNA: 00-00-00-00-00-00.
 02:31:14	NIF	UNA change at station 08-00-2B-3B-FA-2D OLD_UNA: 
			00-00-00-00-00-00  NEW_UNA: 00-80-16-08-80-9E.
 02:31:19	MAC	Ring NOT operational.                                 
 02:31:19	MAC	Ring operational.                                     
 02:31:20	MAC	"Detected 27 CLAIM frame(s) in last second, OLD_TNEG: 
			5000 uS, NEW_TNEG: 5000 uS."
 02:32:07	SIF	"Station 08-00-2B-BB-EF-B7 RsrcIdx 2,MACError-Ct 
			changed in 60 sec, OLD_VAL:1, NEW_VAL:2"
 02:33:07	Int	Detected 1 IFG(s) less than 6 bytes in last second.   
 04:15:45	Int	Detected 1 IFG(s) less than 6 bytes in last second.   


Malcolm
1594.2Slight change in descriptionBUNDE::BURNSMon Feb 20 1995 11:2089
In the base note, the description of the situation included the statement:

"In another test, when an new SAS was inserted and powered on and, somewhat 
later, off, events of the "chaotic" type described above were produced."

It turns out that a record of ring behavior during a period when a VAX was 
having its FDDI adapter replaced is available from the W&G analyser.  
This record, included below, shows a much more orderly behavior than the 
customer had described to me for the SAS insertion test in question.

In the following record, station 08-00-2B-33-5C-BA is eventually replaced 
by station 08-00-2B-36-D0-E2.  Unfortunately, I can't describe the steps 
actually taken by the MCS engineer during this period.


___________________________________
 05:32:17	SIF	"Station AA-00-04-00-E6-0F RsrcIdx 1,PORTLem-Ct 
			changed in 59 sec, OLD_VAL:127, NEW_VAL:128"
 06:03:37	NIF	UNA change at station 08-00-2B-35-9A-2E 
			OLD_UNA: 08-00-2B-33-5C-BA  NEW_UNA: 00-00-00-00-00-00.
 06:04:40	MAC	Ring NOT operational.                                  
 06:04:40	MAC	Ring operational.                                      
 06:04:40	MAC	Ring NOT operational.                                  
 06:04:40	MAC	Ring operational.                                      
 06:04:41	NIF	UNA change at station 08-00-2B-35-9A-2E 
			OLD_UNA: 00-00-00-00-00-00  NEW_UNA: 08-00-2B-37-24-44.
 06:04:41	MAC	"Detected 73 CLAIM frame(s) in last second, 
			OLD_TNEG: 5000 uS, NEW_TNEG: 5000 uS."
 06:05:41	NIF	"Station 08-00-2B-33-5C-BA deleted from ring map, 
			no NIF."
 06:05:52	MAC	Ring NOT operational.                                  
 06:05:52	MAC	Ring operational.                                      
 06:05:53	NIF	Station 08-00-2B-33-5C-BA detected on ring.            
 06:05:53	NIF	UNA change at station 08-00-2B-35-9A-2E 
			OLD_UNA: 08-00-2B-37-24-44  NEW_UNA: 08-00-2B-33-5C-BA.
 06:05:53	MAC	"Detected 27 CLAIM frame(s) in last second, 
			OLD_TNEG: 5000 uS, NEW_TNEG: 5000 uS."
 06:06:17	SIF	"Station 08-00-2B-35-9A-2E RsrcIdx 2,MACError-Ct 
			changed in 60 sec, OLD_VAL:0, NEW_VAL:1"
 06:13:57	MAC	Ring NOT operational.                                  
 06:13:57	MAC	Ring operational.                                      
 06:13:57	MAC	Ring NOT operational.                                  
 06:13:57	MAC	Ring operational.                                      
 06:13:58	NIF	UNA change at station 08-00-2B-35-9A-2E 
			OLD_UNA: 08-00-2B-33-5C-BA  NEW_UNA: 08-00-2B-37-24-44.
 06:13:58	MAC	"Detected 73 CLAIM frame(s) in last second, 
			OLD_TNEG: 5000 uS, NEW_TNEG: 5000 uS."
 06:19:01	NIF	"Station 08-00-2B-33-5C-BA deleted from ring map, 
			no NIF."
 06:31:17	SIF	"Station AA-00-04-00-E7-0F RsrcIdx 1,PORTLem-Ct 
			changed in 59 sec, OLD_VAL:964, NEW_VAL:965"
 06:39:59	MAC	Ring NOT operational.                                  
 06:39:59	MAC	Ring operational.                                      
 06:39:59	SRF	PORTPathChange Event at Station 00-00-1D-10-64-6D 
			PORT Index = 10.
 06:40:00	NIF	Station 08-00-2B-36-D0-E2 detected on ring.
 06:40:00	NIF	UNA change at station 08-00-2B-35-9A-2E 
			OLD_UNA: 08-00-2B-37-24-44  NEW_UNA: 08-00-2B-36-D0-E2.
 06:40:00	MAC	"Detected 27 CLAIM frame(s) in last second, 
			OLD_TNEG: 5000 uS, NEW_TNEG: 5000 uS."
 06:40:17	SIF	"Station 08-00-2B-35-9A-2E RsrcIdx 2,MACError-Ct 
			changed in 60 sec, OLD_VAL:1, NEW_VAL:2"
 06:45:31	MAC	Ring NOT operational.                                  
 06:45:31	MAC	Ring operational.                                      
 06:45:31	MAC	Ring NOT operational.                                  
 06:45:31	MAC	Ring operational.                                      
 06:45:31	MAC	"Detected 73 CLAIM frame(s) in last second, 
			OLD_TNEG: 5000 uS, NEW_TNEG: 5000 uS."
 06:45:32	NIF	UNA change at station 08-00-2B-35-9A-2E 
			OLD_UNA: 08-00-2B-36-D0-E2  NEW_UNA: 08-00-2B-37-24-44.
 06:46:41	MAC	Ring NOT operational.                                  
 06:46:41	MAC	Ring operational.                                      
 06:46:42	NIF	UNA change at station 08-00-2B-35-9A-2E 
			OLD_UNA: 08-00-2B-37-24-44  NEW_UNA: 08-00-2B-36-D0-E2.
 06:46:42	MAC	"Detected 27 CLAIM frame(s) in last second, 
			OLD_TNEG: 5000 uS, NEW_TNEG: 5000 uS."
 06:47:17	SIF	"Station 08-00-2B-35-9A-2E RsrcIdx 2,MACError-Ct 
			changed in 59 sec, OLD_VAL:2, NEW_VAL:3"
 06:47:36	MAC	Ring NOT operational.                                  
 06:47:36	MAC	Ring operational.                                      
 06:47:37	MAC	"Detected 55 CLAIM frame(s) in last second, 
			OLD_TNEG: 5000 uS, NEW_TNEG: 5000 uS."
 09:05:17	SIF	"Station AA-00-04-00-E6-0F RsrcIdx 1,PORTLem-Ct 
			changed in 60 sec, OLD_VAL:128, NEW_VAL:129"



Malcolm