Enter your search terms Submit search form
	Web pdp-10.trailing-edge.com

      ---------------
      |d|i|g|i|t|a|l|                   INTEROFFICE MEMORANDUM
      ---------------
      
      
      TO: DECnet Users                  DATE: 4-Mar-80
                                        FROM: NCSS
                                        EXT: HOTLINE (6492)
                                        LOC/MAIL STOP: MR1-2/H22



      SUBJ:  The use of 32K DN20's






           DECnet-20 V3.0 WILL NOT RUN in a  32K  DN20.   The  following
      information describes DECnet-20 V2.0 or V2.1.

           The 32K DN20, supporting one DUP, is  very  short  on  buffer
      space.   A  pontential  reliability problem exists because of this
      shortage.  If the DUP is using a noisy or  moderately  noisy  line
      then  some  buffers  may  have  to be retained for retransmission.
      When this happens, buffer space can get so low that the  MCB  will
      automatically  shut  the line down and allow it to restart after a
      few moments.  When this problem occurs The operator  will  see  at
      the console:

      Node BLAH Offline
      Node BLAH Online

      The user may see his link hang, or he may get an IO error message.

           The problem can be solved by lowering the demand on  buffers.
      This can be done using any combination of the following.

           1) Use a line with fewer errors.

           2) Lower the number of buffers sitting in the MCB by  setting
      the  flow  control  count  in the remote node to fewer messages or
      segments.  If the remote node  is  a  TOPS20  system  see  3)  for
      patches.

           3) Lower the number of buffers sitting in the MCB by  setting
      the  flow  control count in the local node lower.  In NSPSRV patch
      as follows where 5 and 3 are smaller than 11 and  10.   Experiment
      with smaller values if necessary.

          $GET SYSTEM:MONITR.EXE
          $START 140
          NSPSRV$:
          STMXDF/  MOVEI T2,11  MOVEI T2,5

                                                                  Page 2


          STMXDF+3/  MOVEI T2,10        MOVEI T2,3
          ^Z
          $SAVE


           4) Lower the number of buffers sitting in the MCB by limiting
      the  number  of  messages queued to the DTE at the local node.  In
      NSPSRV patch as follows:

          $get system:MONITR.EXE
          $ST 140
          NSPSRV$:
          STMXDF+5/ MOVEI T2,10 MOVEI T2,3
          ^Z
          $SAVE