Conference 7.286::fddi

532.0. "Problem with DEMFA if MAXBUF=2048." by VCSESU::WADE (Bill Wade, VAXc Systems & Support Eng) Wed Apr 08 1992 12:08

    
    I thought this note from the DECnetVAX note conference should be
    entered in the FDDI conf as well.  It describes a potential problem
    with a DEMFA on a system that has MAXBUF set to the default value of
    2048.
    
    /bill
           <<< BULEAN::BULEAN$USER:[NOTES$LIBRARY]DECNETVAX.NOTE;1 >>>
                           -< DECnet-VAX Conference >-
================================================================================
Note 4397.0                  Line synch lost problem                   4 replies
30380::WADE "Bill Wade, VAXc Systems & Support Eng"  22 lines   2-APR-1992 13:28
--------------------------------------------------------------------------------
    I'm working on a customer problem with a bouncing DECnet circuit.
    
    They have a VAX 6000 running VMS 5.4-3, configured as a level 1 
    router equipped with a DEMNA and a DEMFA.  
    When they turn the DEMFA circuit on it goes up and down with
    line synch lost events.  This causes havoc with the other routers on
    the net.
    
    At first look they have the line receive buffers set to 8 on the DEMFA
    which seems too low.  This looks like a possible problem.
    They have another 6000 (VMS 5.5) on the net with the same configuration 
    and all is fine. The line receive buffers are set to 32 on its  DEMFA. 
    
    I was told that even when they isolate this system from the
    network and bring up the DEMFA circuit it still bounces so it doesn't
    look like a problem due to not enough receive buffers.  
    
    They have already swapped the DEMFA.  Anybody have any insight into this 
    problem?
     
    Thanks,
    Bill
================================================================================
Note 4397.1                  Line synch lost problem                      1 of 4
CSC32::B_GOODWIN "Time is an illusion..."             2 lines   2-APR-1992 14:58
--------------------------------------------------------------------------------
How about providing some line counters from the MFA-0 line? Maybe their is 
something phyically wrong.
================================================================================
Note 4397.2                  Line synch lost problem                      2 of 4
STAR::SALKEWICZ "It missed... therefore, I am "      16 lines   2-APR-1992 15:15
                              -< help on the way >-
--------------------------------------------------------------------------------
    FYI,.. I'm working with Bill off linbe to resolve the CLD,..
    
    Howevere,. I will repost in the next reply a very handy debugging
    SDA script with informative text on how to use it. It does require
    that you have access to some symbol tables FXSYM.STB for the correct
    release of VMS. Getting access to the correct symbol tables is really
    a matter of getting access to VMS results disks. If you have the need
    to do this, please establish a reliationship with someone in your
    management, or VMS management who can open that door for you. Support
    people who look at problems in the field should have no trouble
    convincing VMS amnagement that they need access to these disks.
    
    In any event,.. see the next reply...
    
    								/Bill
           
    
================================================================================
Note 4397.3                  Line synch lost problem                      3 of 4
STAR::SALKEWICZ "It missed... therefore, I am "     532 lines   2-APR-1992 15:16
                         -< the SDA scripts and text >-
--------------------------------------------------------------------------------
    

    .
    .
    .
    Script deleted.  Look in decnetvax notes conference.
    .
    .
    .
    
    

================================================================================
Note 4397.4                  Line synch lost problem                      4 of 4
30380::WADE "Bill Wade, VAXc Systems & Support Eng"  56 lines   8-APR-1992 11:32
                      -< Problem was with MAXBUF=2048.  >-
--------------------------------------------------------------------------------
    After further investigation Bob Lyons and I found that the customer had 
    modified some of their sysgen parameters  and evidently had not written 
    them to the current param file.
    
    
	A quick overview of the problem:

	After a VAXcluster reboot DECnet on the two 6000 systems in the 
    	VAXcluster would bounce the MFA-0 (DEMFA) circuit about every 10 sec. 

	%%%%%%%%%%%  OPCOM   8-APR-1992 08:39:04.64  %%%%%%%%%%%    
	(from node FNALA  at  8-APR-1992 08:39:04.63)
	Message from user DECNET on FNALA
	DECnet event 4.10, circuit up
	From node 42.2 (FNALA),  8-APR-1992 08:39:03.09
	Circuit MFA-0

	%%%%%%%%%%%  OPCOM   8-APR-1992 08:39:10.20  %%%%%%%%%%%    
	(from node FNALA  at  8-APR-1992 08:39:10.18)
	Message from user DECNET on FNALA
	DECnet event 4.7, circuit down, circuit fault
	From node 42.2 (FNALA),  8-APR-1992 08:39:09.31
	Circuit MFA-0, Line synchronization lost

	%%%%%%%%%%%  OPCOM   8-APR-1992 08:39:36.48  %%%%%%%%%%%    
	(from node FNALA  at  8-APR-1992 08:39:36.46)
	Message from user DECNET on FNALA
	DECnet event 4.10, circuit up
	From node 42.2 (FNALA),  8-APR-1992 08:39:36.43
	Circuit MFA-0
	
	%%%%%%%%%%%  OPCOM   8-APR-1992 08:39:40.67  %%%%%%%%%%%    
	(from node FNALA  at  8-APR-1992 08:39:40.66)
	Message from user DECNET on FNALA
	DECnet event 4.7, circuit down, circuit fault
	From node 42.2 (FNALA),  8-APR-1992 08:39:38.08
	Circuit MFA-0, Line synchronization lost

    After some thought the customer realized that some of there  sysgen 
    parameters had been modified and evidently had not been written to the 
    current param file prior to the reboot.   
    
    I compared the sysgen params with a 6000/DEMFA that was running fine 
    in another Fermi VAXcluster and found that MAXBUF was set to 8196 on the 
    good system and 2048 on the bad.  

    After setting MAXBUF to 8196 the circuit came up and all is fine.   
    Looks like a MAXBUF value of ~<4500 is a no go with the DEMFA. Bob Lyons 
    and I will do what needs to be done to get this documented.
                              
    A special thanks to B. Salkewicz for getting his DEMFA diagnostic command 
    procedures to us.  They proved that the DEMFA was fine and the problem
    was somewhere between DECnet and the DEMFA.

    /bill