voss@cs.uiuc.edu (Bill Voss) (04/16/91)
I just started running xntpd version 1.3 on my SparcStation last week. It runs as a stratum 3 client of the campus's three stratum 2 servers. Each of the campus's stratum 2 servers is peer with a different pair of off-campus stratum 1 servers, and also the other two campus servers. At this instant, I have an xclock -digital -update 1 running on one of the ntp servers, and another on my local workstation. To my vast surprise, the two xclock's disagree by roughly 15 seconds. (Unfortunately I don't have accounts on the other two servers.) Is this normal? I expected them to agree to within a fraction of a second, not disagree by 1/4 minute. Would someone with accounts on multiple stratum 2 servers please repeat my simple experiment? thanks, Bill Voss <voss@cs.uiuc.edu> P.S. The large delay is because I'm operating over a slip connection, but even that delay is only a fraction of a second. laslo$ ntpq -p ; xntpdc -s remote refid st when poll reach delay offset disp ========================================================================== +uxc.cso.uiuc.ed 132.249.16.1 2 1011 1024 377 301.1 -8.47 446.2 +ux2.cso.uiuc.ed 192.43.244.9 2 521 1024 377 300.8 9.09 7.4 *garcon.cso.uiuc 128.8.10.1 2 11 64 377 301.5 4.33 5.8 remote local st poll reach delay offset disp ====================================================================== +ux2.cso.uiuc.ed 128.174.5.76 2 1024 377 0.3008 0.009090 0.0074 +uxc.cso.uiuc.ed 128.174.5.76 2 1024 377 0.3011 -0.008467 0.4462 *garcon.cso.uiuc 128.174.5.76 2 64 377 0.3015 0.004333 0.0058 laslo$