[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

1150.0. "Port LED goes RED indicating LCT failure" by CUJO::HILL (Dan Hill-Net.Mgt.-Customer Resident) Tue Nov 16 1993 15:23

    We have a customer running V3.2 firmware on a DC500 to which 3 Sun
    workstations are connected.  One of the Suns seems to have a problem as
    it will send the its port on the DC500 into a fault state.  The port
    LED goes RED (not flashing), indicating a Link Confidence Test (LCT)
    failure.  At this point, all other port LEDs go RED.  
    
    The workaround involves unplugging each port's cable and plugging it
    back into the port.
    
    We are in the process of scheduling time to perform loopback tests and
    to swap FDDI cables/cards, if necessary.
    
    The customer wants to know what is going on before they go to the
    trouble of swapping everything out.  They don't want to do all of this
    only to find that the problem is elsewhere.
    
    Could it be that the "bad Sun" is propagating a corrupted frame on the
    ring?
    
    Thanks,
    Dan
T.RTitleUserPersonal
Name		DateLines
1150.1KONING::KONINGPaul Koning, B-16504Tue Nov 16 1993 16:209
It's possible for a station to send a frame that sets off the LCT.  The more
common cause is a bad cable or transceivers, of course.  But LCT is designed
in such a way that the symptoms of trouble are isolated to one connection.
Any error pattern that LCT reacts to is NOT propagated to other ports; any
data pattern that does propagate is one that LCT does not object to.  So
a situation where a problem on one port causes all the others to go into a
bad state sounds like a concentrator problem.

	paul
1150.2Concentrator has been replaced alreadyCUJO::HILLDan Hill-Net.Mgt.-Customer ResidentWed Nov 17 1993 09:1214
    Paul,
    Thanks for the prompt reply.  The "bad Sun" was moved numerous times to
    different port cards and the problem followed it.  Finally, the entire
    concentrator was replaced.  The problem still exists.  That doesn't
    mean that the replacement concentrator was a good one, so I'll
    recommend that yet another concentrator be put in place if this happens
    again.
    
    As a side note, the "bad Sun's" FDDI controller was replaced a couple
    of days ago.  So far, no problems.  Fiber cables have not been
    replaced.\
    
    Thanks,
    Dan
1150.3KONING::KONINGPaul Koning, B-16504Wed Nov 17 1993 15:2412
Strange.

Note that it is certainly possible (and given how things "follow the Sun"
extremely likely) that the original problem where one port goes red is the
Sun's fault.  In general, an LEM/LCT error can be caused by either endpoint
of that connection, or the cabling in between.

But that doesn't explain the other ports going bad as well.  There may be
a bug somewhere that's triggered by some strange condition, and fixing the
original trigger made everything work...

	paul
1150.4LCT / card problem went away with V3.3CUJO::HILLDan Hill-Net.Mgt.-Customer ResidentThu Dec 02 1993 21:2110
    We upgraded the firmware to V3.3 and the problem went away.  Weird,
    because, from what Paul tells us, there was nothing in V3.3 that should
    have fixed such a problem.  
    
    Does anyone have the name of the DEFCN (DECconc500) product manager?
    
    I'd like to have that group pursue this further.
    
    Thanks,
    Dan
1150.5QUIVER::STEFANIHave the # for the Mars Observer?Thu Dec 02 1993 21:466
    >>Does anyone have the name of the DEFCN (DECconc500) product manager?
    
    I believe Sharon Oneill (DELNI::ONEILL) is the product manager for all
    of the FDDI concentrators.
    
       - Larry
1150.6what frames causes LCT?NPSS::TAYLORWed Jun 01 1994 15:289
    
    re: .1
    
    Paul, 
    
    	What type of frame could set of the LCT?
    
    wayne
1150.7KONING::KONINGPaul Koning, B-16504Wed Jun 01 1994 17:1013
There's no way for software to generate frames that trigger LCT, if the hardware
is designed with even modest amount of competence.  It would take a really
amazing hardware bug to make this possible.  

Basically, LCT triggers on bit patterns that will not be sent by a conforming
PHY chip.  Since bit errors on the medium can produce all possible bit patterns,
the arrival of invalid patterns is an indication of bit errors.  To first
approximation, a legal bit pattern on the wire that is hit by a bit error has
roughly a 33% change of being turned into an illegal bit pattern (the other
alternative being a legal but different bit pattern, something that the CRC 
check in the receiving station will see but LCT will not).

	paul