My sparc-20 running -current (2001/03/24) is having trouble keeping
accurate time.  For some reason ntpd wants to run about 2-5 seconds
offset from my local server.  (The local time server is my NetBSD/sparc
1.3.2 SS2.)

It syncs up OK and then very quickly drifts off.  None of the other
clients have any problems staying in sync (though the other day one
really ancient BSD/OS 1.1 machine suddenly drifted about 4 seconds and
then its xntpd complained about too large an offset and died, though the
actual system clock was less than a second off when I reset it later
with ntpdate).  My i386s running not-so-current (2000/09/21) are
included in the list of the clients not having any problems.

I do have "options NTP" in my kernel (basically a GENERIC clone).

Here's an example of restarting it, it having drifted off when I check
it about 20 minutes later, and some other possibly useful information:

03:07 [1093] # ntpdc
ntpdc> dmpeers
     remote           local      st poll reach  delay   offset    disp
=======================================================================
*most.weird.com  204.92.254.18    2   64  377 0.00226  4.284553 0.00093
ntpdc> 03:08 [1094] # 
03:08 [1094] # 
03:08 [1094] # /etc/rc.d/ntpd stop
Stopping ntpd.
03:08 [1095] # /etc/rc.d/ntpdate start
Setting date via ntp.
03:08 [1096] # /etc/rc.d/ntpd start    
Starting ntpd.
03:08 [1097] # ntpdc           
ntpdc> dmpeers
     remote           local      st poll reach  delay   offset    disp
=======================================================================
 most.weird.com  0.0.0.5         16   64    0 0.00000  0.000000 0.00000
03:08 [1098] # ntpdc  
ntpdc> dmpeers
     remote           local      st poll reach  delay   offset    disp
=======================================================================
 most.weird.com  204.92.254.18    2   64    1 0.00211 -0.010670 7.93750
ntpdc> 03:09 [1099] #
03:29 [1099] # ntpdc
ntpdc> dmpeers
     remote           local      st poll reach  delay   offset    disp
=======================================================================
*most.weird.com  204.92.254.18    2   64  377 0.00301  4.626375 0.00096
ntpdc> showpeer most.weird.com
remote 204.92.254.2, local 204.92.254.18
hmode client, pmode mode#255, stratum 2, precision -16
leap 00, refid [204.152.184.72], rootdistance 0.13324, rootdispersion 0.01877
ppoll 6, hpoll 6, keyid 0, version 4, association 22060
valid 7, reach 377, unreach 0, flash 0x0000, boffset 0.00400, ttl/mode 0
timer 0s, flags system_peer, config, bclient
reference time:      be6ec287.0073a000  Fri, Mar 30 2001  3:29:59.001
originate timestamp: be6ec2e6.f75f1000  Fri, Mar 30 2001  3:31:34.966
receive timestamp:   be6ec2df.e60c282c  Fri, Mar 30 2001  3:31:27.898
transmit timestamp:  be6ec2df.e5422ccb  Fri, Mar 30 2001  3:31:27.895
filter delay:  0.00238  0.00374  0.00230  0.00264 
               0.00365  0.00264  0.00233  0.00237 
filter offset: 7.068867 6.700055 6.330232 5.960622
               5.591561 5.221984 4.653440 4.283908
filter order:  0        1        2        3       
               4        5        6        7       
offset 6.264424, delay 0.00272, error bound 0.03081, filter error 0.31401
ntpdc> sysinfo
system peer:          0.0.0.0
system peer mode:     unspec
leap indicator:       00
stratum:              16
precision:            -18
root distance:        0.13579 s
root dispersion:      0.44331 s
reference ID:         [204.92.254.2]
reference time:       be6ec31f.e69ee45c  Fri, Mar 30 2001  3:32:31.900
system flags:         auth monitor ntp kernel stats kernel_sync 
jitter:               0.312790 s
stability:            6.000 ppm
broadcastdelay:       0.003998 s
authdelay:            0.000000 s
ntpdc> ^D03:30 [1100] #
03:30 [1100] # cat /var/db/ntp.drift
500.000
03:30 [1101] # 

I'm going to add another peer and see how it fairs....

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>