[comp.protocols.time.ntp] One works, One fails

ramus@WINDSAIL.NERSC.GOV (Joe Ramus) (03/20/91)

We have two SUN Sparcs running the same xntp software.
One machine gets a response from  fuzz.sdsc.edu and the other
gets nothing.
Both machines can use "ntpq" to query the status of fuzz.sdsc.edu.
-------------------------------------------------------------------------
This machine seems to get a response from  fuzz.sdsc.edu

{rick:2} /usr/local/bin/ntpq -c peers windsail.nersc.gov
     remote           refid      st  when poll reach  delay  offset   disp
==========================================================================
+fuzz.sdsc.edu   .WWVB.           1   994 1024  377    80.5   -5.20    5.8
*clepsydra.dec.c .GOES.           1    30   64  377    54.6   -1.20    3.6
+little-walter.C ncarfuzz.ucar.e  2   711 1024  377   100.6   -9.36    8.1
+polaris.llnl.go norad.arc.nasa.  2   504 1024  376    54.7  -10.65    3.5
 windmill.nersc. fuzz.sdsc.edu    2   560 1024  376    31.1   -6.29    1.6
 rick.nersc.gov  windsail.nersc.  3    63   64  376    38.7    4.59    0.2

-------------------------------------------------------------------------
This machine gets no response from  fuzz.sdsc.edu
 
{rick:3} /usr/local/bin/ntpq -c peers rick.nersc.gov
     remote           refid      st  when poll reach  delay  offset   disp
=======================algorith===================================================
 fuzz.sdsc.edu   0.0.0.0         16 never 1024    0     0.0    0.00  64000
+fuzz.psc.edu    truechimer.cso.  2   289 1024  377   174.9   -7.06   13.4
+polaris.llnl.go norad.arc.nasa.  2   733 1024  377    70.0   -9.98    4.1
*windsail.nersc. clepsydra.dec.c  2     1   64  377    39.3   -4.32    0.1
 prism.nersc.gov 0.0.0.0         16 never 1024    0     0.0    0.00  64000
+windjammer.ners windsail.nersc.  3   821 1024  376    32.9  -11.40    2.9
+windmill.nersc. fuzz.sdsc.edu    2   866 1024  376    34.9  -12.91    2.4

Mills@udel.edu (03/20/91)

Joe,

See my forlorn wail to this list a few weeks ago. The fuzzball primary
servers now limit service to no more than two hosts from the same net.
This was done to reduce congestion and sustain the design accuracy. You
may need to distribute the load over many servers, rather than gang up
on one. You may also consider peering with only the stratum-2 servers,
which are probably more robust than the primary servers anyway.

If the above does not pertain to you, please bleat.

Dave