Minutes of the 8/16/95 Grex Staff Meeting

Aug 20 16:36 UTC 1995

    Minutes of the 8/16/95 Staff Meeting
    ------------------------------------ 
    The meeting was called to order by Valerie Mates at about
    8 p.m. Present: Valerie Mates (popcorn), Marc Unangst (mju),
    Rob Henderson (robh), Scott Helmke (scott), and John
    Remmers (remmers).
 
    Sun-4 status.
 
       Marc reported on the work he and Greg Cronau have been
       doing on the Sun-4 box. The disks are partitioned.
       The 1.7gb Micropolis will be drive 0. SUN/OS has been
       installed but the file system did not come up
       successfully. This may be a configuration problem;
       Greg expects to work on it later tonight.
         Making certain file systems (e.g. /usr/local) read only 
       to minimize security risks is under consideration.
         Once the basic system is in place, the machine will
       be accessible to staff via telnet for installing software.
       Marc estimated that this will happen in the next day or two.
       We decided to discuss the details of software installation
       online.
         It was mentioned that part of making the machine Grex-
       ready would be to install the kernel patches necessary
       to implement the internet access policies for verifier
       non-memgers.
         Marc estimated that the machine could be ready for Grex
       to run on by mid to late September.
 
    Ttyh9.
 
       We have a modem for it that needs to be installed. Marc
       volunteered to take care of this.
 
    UID blowout.
 
       Valerie indicated that she thought this topic had been
       discussed sufficiently. Nobody disagreed.
 
    Fixing the terminal option in the "change" program.
 
       John indicated interest and will look into this as time
       permits. Marc suggested that rewriting "change" in C (it
       is written in Perl, an interpreted language) would sub-
       stantially improve its performance.
 
    IP packet fragmentation.
 
       Gryps currently breaks outgoing IP packets into fragments
       to optimize interactive performance. The problem is that
       although packet fragments are acceptable according to TCP/IP
       standards, not all routers handle them, the result being
       that Grex can't talk to certain ones, e.g. the Merit NAS.
       Marc outlined three possible solutions:
 
        1. Have Gryps stop doing fragmenting. The drawback to
       this is that interactive performance in telnet sessions
       would take a big hit when large data transfers are taking
       place, e.g. via ftp.
        2. Reduce the size of the packets that Grex sends to
       Gryps, so the Gryps wouldn't have to fragment them.
        3. Bitch at Merit to fix their software. It was pointed
       out that in the best of all possible worlds, Grex shouldn't
       have to devise workarounds for other people's broken
       software.
 
       No definitive conclusions were reached on what to do.
 
    The meeting was declared concluded at about 8:45.

response 2 of 2:

Aug 20 22:17 UTC 1995

I also pointed out a problem with the current version of
the "icant" program.  The .login file it produced for me
combined the stty and tset statements on one line, and
neither one actually worked.  I'd like to see this fixed,
but I don't know if I'm brave enough to even look at
the source code, from what others tell me.  >8)