[comp.sys.apollo] SR10.3 and diskless nodes

gjalt@ele.tue.nl (Gjalt de Jong) (03/07/91)

When we run SR10.3 on a diskless machine (or when we've booted a disked
machine on a SR10.3 node), it can't find the registry and is therefore using
the localhistory. This is not the case for the disked 10.3 nodes.
Also is doesn't depend whether llbd is running or not.

Does anyone knows the magic of this?
--
__
Gjalt G. de Jong,                 | Phone: +(31)40-473345
Eindhoven University of Technology, Dept. of Electr. Eng. (ES/EH 7.26)
P.O. Box 513, 5600 MB Eindhoven, The Netherlands
Email: gjalt@es.ele.tue.nl

hanche@imf.unit.no (Harald Hanche-Olsen) (03/08/91)

In article <GJALT.91Mar7090953@eutes3.ele.tue.nl> gjalt@ele.tue.nl (Gjalt de Jong) writes:

   When we run SR10.3 on a diskless machine (or when we've booted a disked
   machine on a SR10.3 node), it can't find the registry and is therefore using
   the localhistory. This is not the case for the disked 10.3 nodes.
   Also is doesn't depend whether llbd is running or not.

   Does anyone knows the magic of this?

Magic is the word.  This behaviour is known (to me) under all
releases: SR10.{0,1,2,3}.  Thanks to helpful people in this forum, we
made sure to run glbd and rgyd on the same node(s).  This helped a
lot, but is not a universal cure.  Every once in a while (it's been a
couple months now) when things get too bad we reboot the entire
network, starting with the glbd/rgyd node.  The order is important;
use the most stable node you can find for running these daemons.

- Harald Hanche-Olsen <hanche@imf.unit.no>
  Division of Mathematical Sciences
  The Norwegian Institute of Technology
  N-7034 Trondheim, NORWAY