Date: Fri, 20 Sep 2002 10:06:24 -0500 From: Scott Bolte <listS+freebsd-questions@niss.com> To: questions@FreeBSD.ORG Subject: ntpd running, clock still drifting Message-ID: <200209201506.g8KF6O94002329@crag.niss.com>
next in thread | raw e-mail | index | archive | help
I recently switched from running ntpdate via cron to ntpd. While ntpd starts and runs, the clock drifts badly (over 6 seconds) in just a few hours. It looks to me like ntpd is not doing anything. Summary of what I have: + I'm using 4 publicly available stratum 2 servers. + I have a drift file, which consistently says '0.000'. + I'm trying to gather statistics but nothing is ever written to the files (which I touched by hand). + I've followed the ntp hints for freebsd and ensured my kernel has POSIX scheduling calls. + I've tried running ntpd with one or more -d options and allowed it to run for nearly a day. In that case, and when I allow it to run in the background, I never see a message that it is setting the clock. + I am running FreeBSD 4.6-RELEASE. + ntpdate does work, so I doubt there are any network/firewall issues. In the 55 minutes since I rebooted with a kernel with _KPOSIX_VERSION=199309L the clock has drifted -.48 seconds. Any help would be appreciated. Details included below. Thank you, Scott ----- Detailed information ------- crag% cat /etc/ntp.conf # # FreeBSD NTP help: # http://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/ntp.html # file://localhost/usr/share/doc/ntp/index.htm # # Server list from http://www.eecis.udel.edu/~mills/ntp/servers.htm # and http://www.eecis.udel.edu/~mills/ntp/clock2.htm # server ntp3.cs.wisc.edu prefer server ntp-2.cso.uiuc.edu server ntp1.cs.wisc.edu server ntp-1.cso.uiuc.edu driftfile /var/db/ntp.drift statistics clockstats peerstats loopstats statsdir /var/log/ntpstats/ filegen peerstats file peerstats type day enable filegen loopstats file loopstats type day enable filegen clockstats file clockstats type day enable restrict default ignore # IP address sanitized restrict X.X.X.X mask 255.255.255.0 notrust nomodify notrap crag% grep ntp /etc/rc.conf ntpdate_enable="YES" ntpdate_flags="-b ntp3.cs.wisc.edu ntp-2.cso.uiuc.edu ntp1.cs.wisc.edu" xntpd_enable="YES" crag% cat /var/db/ntp.drift 0.000 crag% ls -l /var/log/ntpstats/ total 0 -rw-r--r-- 1 root wheel 0 Sep 19 18:04 clockstats -rw-r--r-- 1 root wheel 0 Sep 19 18:04 loopstats -rw-r--r-- 1 root wheel 0 Sep 19 18:04 peerstats crag% cat /usr/src/contrib/ntp/html/hints/freebsd If you are compiling under FreeBSD and see messages in the syslogs that indicate that the ntpd process is trying to use unavailable sched_ calls, it means you are running a kernel that does not have the POSIX scheduling calls enabled. You have two choices: - Ignore the messages - Generate a new kernel, where the kernel configuration file contains the lines: options "P1003_1B" options "_KPOSIX_PRIORITY_SCHEDULING" options "_KPOSIX_VERSION=199309L" crag% strings /kernel | egrep '^___' | egrep '_KPOSIX|P10' ___options P1003_1B #Posix P1003_1B real-time extensions ___options _KPOSIX_PRIORITY_SCHEDULING ___options _KPOSIX_VERSION=199309L # needed by ntpd? crag% tail /var/log/messages | grep ntp Sep 20 09:05:03 crag ntpd[125]: ntpd 4.1.0-a Tue Jun 11 03:57:50 GMT 2002 (1) Sep 20 09:05:03 crag ntpd[125]: kernel time discipline status 2040 crag% uptime 9:59AM up 55 mins, 8 users, load averages: 0.00, 0.00, 0.00 crag% ntpdate -d ntp3.cs.wisc.edu ntp-2.cso.uiuc.edu ntp1.cs.wisc.edu |& tail -1 20 Sep 09:59:28 ntpdate[2191]: adjust time server 128.105.39.11 offset -0.477759 sec To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200209201506.g8KF6O94002329>