IMP System change notification

Network Working Group                                        A. McKenzie
Request for Comments: 343                                        BBN-NET
NIC #10422                                                   19 May 1972
Category: B.1
Obsoletes:  RFC #331


                     IMP System Change Notification

   As most of you probably know, our attempt to release IMPSYS 2600 on
   May 15 (as discussed in RFC #331) was unsuccessful.  Briefly, the
   primary reasons for failure (all operational rather than programming
   bugs) were:

      1) The network appears to have grown too large for us to
         successfully coordinate operator intervention at all sites.

      2) An unanticipated aspect of the gradual release of the new
         system was the "poisoning" of IMP's running the new system by
         adjacent IMP's running the old system, and vice versa.  Working
         around this problem made the dissemination of the new system
         take much longer than anticipated.

   We believe we see solutions to both of these problems and are now
   ready to schedule the next attempt(s) to release IMPSYS 2600.  We
   hereby request all sites to cooperate with this effort in the
   following ways.

I.   All sites with Model 516 IMP's have been contacted by telephone and
     asked to set the switches on their IMP's as follows:

        HALT INH     - off
        W.D.T.       - off
        AUTO RSTRT   - on
        MEMORY PRTCT - off

        Please be sure that your switches remain set at these positions
        until further notice.

II.  No site should attempt to use the IMP's "Trace" routine during the
     day(s) of the attempted system release.

III. Terminal IMP's may be turned off for significant portions of the
     day(s) of the attempted system release.  Please be patient with any
     inconvenience which this causes.

IV.  We hope to accomplish release of IMPSYS 2600 on the next attempt.
     Nevertheless, we are not omniscient and may again be thwarted by
     difficulties.  Accordingly, we are scheduling the next attempt for
     Wednesday, May 31, and subsequently every Tuesday (June 6, 13, 20,
     ...) until the release is successfully accomplished.  We will
     arrange the release process so that site operator coverage is not
     required; nevertheless we hope that the normal site operator
     coverage is available in case unanticipated problems develop.
     Release of the new system will begin early in the day
     (approximately 7 a.m. EDT).  We again suggest that sites do not
     schedule production work over the network during the release
     attempt(s).  We will notify each site by phone after the release
     has been successfully completed.

     Finally, we apologize in advance for any inconvenience caused by
     the contortions required to release this new system, and thank all
     of the site operations' staffs who worked with us on May 15 for
     their cooperation.

     AAM/jm












          [ This RFC was put into machine readable form for entry ]
          [ into the online RFC archives by Ronald de Leeuw 1/99  ]