[rescue] ss2 under load

jwbirdsa at picarefy.com jwbirdsa at picarefy.com
Fri Feb 15 17:15:07 CST 2002


>Penn State, circa 1990, VAX 11/750 running BSD, load average of between 100 and
>150 depending on what part of the day it was.  it worked just as well then as
>when the load average was 2.  seriously.  slow, but consistantly slow.
>
>-brian

   Princeton University, 1986-1990, VAX 8600, rumored to be dual-processor,
running Ultrix. Became unusably slow with about 50 users logged on, usually
crashed if it hit 70. The problem was self-limiting because once it reached
the danger zone, it was normally almost impossible for more users to log in:
the one-minute timeout would fire before login finished collecting all the
characters in your password.

   It was replaced with a Sun, IIRC a VME Sparc (4/3xx? 4/4xx? 4/6xx?), which
completely toasted its butt.

   Did we just have a lemon? Was Ultrix that bad? I admit I have a soft
spot in my heart for VAXen because that 8600 was my first Unix, and I have
several in my collection (MVII, Vaxserver 3100-??, 4000VLC) which I hope to
put into service someday, but I have never thought of them as being fast.
At my first workplace we had an 11/780 NFS-serving a couple racks of
Fujitsu Eagles. Trying to use it interactively was agonizing -- the Sun
3/60 on my desk ran rings around it -- but I suspect that it was getting
cratered by NFS traffic from a hundred or so workstations, so it had some
justification for sluggish response.

   --James B.



More information about the rescue mailing list