jmattson@UCSD.EDU (Jim Mattson) (05/04/91)
On a machine listening to broadcast ntp, where the broadcast delay is set to 0.06 seconds, xntpdc sysinfo shows: broadcast delay: 0.000000915490 (On the same machine, if no broadcast delay is specified, xntpdc sysinfo shows the expected default: broadcast delay: 0.007999999914) It seems that the broadcast delay parameter should be handled the same way as the authdelay parameter, so I took the code from there, and now it seems to work. --jim BTW, I chose 0.06 seconds because that was the delay xntpdc reported when I was running as a peer rather than a broadcast client. Is this value meaningful? It seems odd that it's an order of magnitude higher than the default value. *** /tmp/RCSAa09736 Sat May 4 00:11:08 1991 --- ntp_config.c Sun Apr 28 16:05:03 1991 *************** *** 394,400 **** case 'r': do { l_fp tmp; - u_fp utmp; if (!atolfp(optarg, &tmp)) { syslog(LOG_ERR, --- 394,399 ---- *************** *** 405,412 **** "command line broadcast delay value %s is unlikely", optarg); } else { ! utmp = LFPTOFP(&tmp); ! proto_config(PROTO_BROADDELAY, utmp); } } while (0); break; --- 404,410 ---- "command line broadcast delay value %s is unlikely", optarg); } else { ! proto_config(PROTO_BROADDELAY, tmp.l_f); } } while (0); break; *************** *** 734,740 **** case CONFIG_BDELAY: if (ntokens >= 2) { l_fp tmp; - u_fp utmp; if (!atolfp(tokens[1], &tmp)) { syslog(LOG_ERR, --- 732,737 ---- *************** *** 745,752 **** "broadcastdelay value %s is unlikely", tokens[1]); } else { ! utmp = LFPTOFP(&tmp); ! proto_config(PROTO_BROADDELAY, utmp); } } break; --- 742,748 ---- "broadcastdelay value %s is unlikely", tokens[1]); } else { ! proto_config(PROTO_BROADDELAY, tmp.l_f); } } break; *** /tmp/RCSAa09744 Sat May 4 00:11:24 1991 --- ntp_proto.c Sun Apr 28 16:08:03 1991 *************** *** 1921,1927 **** /* * Set default broadcast delay */ ! sys_bdelay = (u_fp)value; break; case PROTO_AUTHENTICATE: --- 1921,1927 ---- /* * Set default broadcast delay */ ! sys_bdelay = (((u_long)value) + 0x00000800) & 0xfffff000; break; case PROTO_AUTHENTICATE:
saf@ATHENA.MIT.EDU (05/09/91)
I have been trying to do a test run for xntpd here at MIT at the Media Lab. I have everything set up. I have root running xntpd and so on. My configuration file for the current test machine is ... peer 18.72.0.3 key 1 # bitsy.mit.edu peer 130.105.1.156 key 2 # clock.osf.org peer 128.118.25.3 key 3 # clock.psu.edu authneticate yes trustedkey 1 2 3 keys /etc/ntp.keys authdelay 0.000323 requestkey 65535 # for mode 7 request driftfile /etc/ntp.drift moniter yes the problem is that I cannot get any updating of the clock when I run it. All I am gettin from ntpq is ... BITSY.MIT.EDU 0.0.0.0 16 never 64 0 0.0 0.00 64000 pmax6.osf.org 0.0.0.0 16 never 64 0 0.0 0.00 64000 otc2.psu.edu 0.0.0.0 16 never 64 0 0.0 0.00 64000 I wonder if you could shed some light on what the problem could be. The drift file is also not being updated. Do I have to add some more configeration or notify the peers before I can get this thing to strart working? I got the software by ftping into louie.udel.edu. Safroadu. Student Researcher At the Media Lab. MIT.