Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!wuarchive!usc!ucsd!ucbvax!WOLF.UWYO.EDU!waldram
From: waldram@WOLF.UWYO.EDU
Newsgroups: comp.sys.apollo
Subject: 10.3 RC TCP after NCS services?
Message-ID: <9101042351.AA05130@wolf.Uwyo.EDU>
Date: 4 Jan 91 23:51:26 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 18

On our DN10000 running SR10.3, we run tcpd, llbd, glbd, and rgyd.  We have had two
occasions where rgyd (a slave) has started utilizing 80%+ of the cpu (dspst).
It has been suggested that if tcpd "blinks" it loses its socket to one of the NCS 
services daemons, but continues to run.  When tcp services are requested, the 
lost socket is under contention.  I do not understand these details.  A suggestion
of starting TCP services (rc.local) after NCS services (just before cron in /etc/rc)
may cure the problem, but may cause other MAJOR problems related to the local site
network setup.  Has anyone:
   1) tried the order switch?
   2) seen more detail on when the switch is safe?
Explanations of the "fuzzy" description/activity above would be appreciated.
           thanks,
               -jjw
Jim Waldram
University of Wyoming
waldram@grizzly.uwyo.edu
jwaldram@outlaw.uwyo.edu
jwaldram@UWYO.BITNET