[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

1433.0. "Can't communicate w/PATHWORKS V4.1 Can't Communicate w/V4.1 PATHWORKS" by DUCAT2::64544::LICAUSE (Al Licause (338-5661)) Tue Sep 03 1991 09:49

I'm running V1.1 DECmcc on a VAXstation 3100 and V4.1 field test of PATHWORKS
for DOS on a DECstation 325 w/DEPCA and 4 MB memory.....NML is loaded in PC.

I can do NCP TELLs all day long, but as soon as I issue any MCC command to
the PC the following results:

NCP>tell four42 sho know cir count


Known Circuit Counters as of  3-SEP-1991 08:41:03

Circuit = ETHER-1

         655  Seconds since last zeroed
           0  Terminating packets received
           0  Originating packets sent
           0  Circuit down
           0  Initialization failure
        6598  Bytes received
        1440  Bytes sent
         114  Data blocks received
          34  Data blocks sent
           0  User buffer unavailable


NCP>tell four42 sho know cir count


Known Circuit Counters as of  3-SEP-1991 08:41:06

Circuit = ETHER-1

         658  Seconds since last zeroed
           0  Terminating packets received
           0  Originating packets sent
           0  Circuit down
           0  Initialization failure
        6779  Bytes received
        1680  Bytes sent
         119  Data blocks received
          40  Data blocks sent
           0  User buffer unavailable




NCP>exit
462> mcc
DECmcc (V1.1.0)

MCC> sho node4 four42 all char

Node4 63.33
AT  3-SEP-1991 08:42:17 Characteristics

Insufficient resources at target entity



Any further attempts to contact the PC via MCC or NCP result in:

	
MCC> exit
462> ncp
NCP>tell four42 sho exec
%NCP-F-CONNEC, unable to connect to listener
-SYSTEM-F-UNREACHABLE, remote node is not currently reachable

NCP>exit
462> mcc
DECmcc (V1.1.0)

MCC> sho node4 four42 line * all count

Node4 four42 Line *
AT  3-SEP-1991 08:49:18 Counters

Node not currently accessible.


Any one have any ideas as to what changed?

I'll also enter this note in the PCSAV2 notes.

Al
                                                                                
T.RTitleUserPersonal
Name
DateLines
1433.1Does this make sense?DUCAT2::64544::LICAUSEAl Licause (338-5661)Tue Sep 03 1991 11:4620
              <<< RANGER::$2$DUA8:[NOTES$LIBRARY]PWDOS4.NOTE;1 >>>
                         -< PATHWORKS for DOS and VMS >-
================================================================================
Note 131.1             Can't communicate w/PATHWORKS V4.1                 1 of 1
PHOTON::Lichtenberg "Mitch Lichtenberg"               6 lines   3-SEP-1991 10:27
--------------------------------------------------------------------------------


It almost sounds like that the MCC programs are trying to make TWO nice
connections to the target node at once.  That probably doesn't work.

/Mitch.



MCC folks....does this make sense.

Hope we don't get into a shouting contest!

Al
1433.2Yep, 2 links (sortof)....TOOK::CAREYThu Sep 05 1991 12:2135
    
    .1 -- you're right on, we are trying to form a second link.
    
    We've been working with the Pathworks folks to iron out the details, 
    but here's a quick look at what goes on:
    
    o We form a link.
    
    o We try to form a second link to find out if the agent can manage
      multiple concurrent links.  If so, we note that.  If not, we note
      that and continue on.
    
    o We make our request.  If its a complex request, and it's possible, we
      run multiple concurrent links to the remote agent in hopes of
      speeding up the response.  For big requests this is a big win;  for
      other requests we're not as sure.  If the remote agent can't handle 
      multiple links, we failover to a sequential approach similar to NCP.
      It requires more patience, but works.
    
    With previous DECnet/DOS incantations, this algorithm worked fine.
    
    Doesn't work with 4.1.  Early indications are that the Pathworks agent
    is getting into some context trouble when we try the second link and it
    doesn't work.  At least, we've been working with the Pathworks folks on
    this one, and they've decided to look at that side first.
    
    What we've seen is our first link succeeds, our second link fails so we
    failover into single link mode.  Since we have a link up, our original
    request usually succeeds (depending on the request), but subsequent
    connection attempts from MCC or NCP fail until the agent is restarted.
    
    More news as it evolves.
    
    -Jim Carey
    
1433.3FYI -- Pathworks folks found problemTOOK::CAREYWed Sep 25 1991 20:0611
    
    Just a quick followup:
    
    We worked with the Pathworks folks, and after some quick conferences
    uncovered some problems in the new Pathworks agent.  I don't have all 
    the details, but there was definitely a context problem in their agent.
    
    The latest word I've gotten is that the problem has been addressed.
    
    -Jim Carey
    
1433.4Update on fix??OTOOA::DOIRON[email protected]Tue Nov 12 1991 16:5614
    Jim,
    
    Could you elaborate on "has been addressed" a little? Is there a patch
    available for pathworks and/or mcc?
    
    To be able to register pathworks v4.1 nodes as node4 entities is
    important when we talk to customer's who have large installed pathworks
    networks in the government space.
    
    It seems more and more situations we are asked to look at require the
    management of large bodies of ps's of some variety.
    
    Thanks,
    Richard
1433.5Pathworks fixing it for their SSB.TOOK::CAREYTue Nov 19 1991 14:4730
    
    Richard,
    
    I have forwarded your request for elaboration over to the Pathworks
    folks.  I don't know if a patch was created, or on what date a fix was
    introduced into the Pathworks field test.
    
    My last discussion with the Pathworks folks was in early October, and
    (like the engineer I am) in inquired whether the problem was understood
    and resolved and the answer was yes.
    
    There won't be any MCC patch.  The PC agent was badly losing context
    after anyone tried multiple connects (even though it doesn't honor
    them, it needs to drop them cleanly) and the Pathworks folks attacked
    it with a vengeance.
    
    Oops, while I was typing this up I just got an update from Pathworks:
    
    	"Yep, we encountered that problem during our field test.
    	Yep, we understand it and have ensured that it will be fixed
    	for our release kit."
    
    I'll send out some feelers about more details, but I don't have any 
    right now.
    
    Hope this helps some.
    
    -Jim Carey