This is a discussion on Synchronize system clock? within the Slackware Linux Support forums, part of the Unix Operating Systems category; --> Le Sun, 17 Dec 2006 18:49:51 +0000, rm a écrit*: > Ivan Rajkovic <ivan@iep.physik.uni-essen.de> wrote: >> On Fri, 15 ...
| |||||||
| FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
| ||||
| Le Sun, 17 Dec 2006 18:49:51 +0000, rm a écrit*: > Ivan Rajkovic <ivan@iep.physik.uni-essen.de> wrote: >> On Fri, 15 Dec 2006 11:59:10 +0100, Niki Kovacs wrote: >> >>> Hi, >>> >>> Is there a simple way to synchronize my system clock, say, on startup? My >>> laptop unfortunately has a broken battery, and everytime I disconnect the >>> power cable, the clock freezes... only to resume activity at the next >>> boot... which always makes me travel back in time >>> >>> Cheers, >>> >>> Niki >> >> put this is /etc/rc.d/rc.local : >> >> /usr/sbin/ntpdate de.pool.ntp.org >> /sbin/hwclock -w >> >> you can change ntp server, here is the list: >> http://ntp.isc.org/bin/view/Servers/WebHome > > We still use rdate. We include the following in rc.local: > > /usr/bin/rdate -s 129.6.15.29 > > Is there a reason to use ntpd over rdate? No special reason for the use you make of this command. It is quite similar to ntpdate, a direct bulk call to get the date/time from NTP server(s). Now, exactly for the same reasons as ntpdate, if you intend to put this command in cron (especially more often than a 24H basis) please consider using ntpd instead. You won't do much harm to the ntp servers for a long time but you might have to find some new ntp servers and/or ISPs ;-) |
| |||
| Loki Harfagr wrote : > Le Sun, 17 Dec 2006 18:49:51 +0000, rm a écrit*: >> Is there a reason to use ntpd over rdate? > > No special reason for the use you make of this command. > It is quite similar to ntpdate, a direct bulk call to > get the date/time from NTP server(s). One think is against rdate and that is the fact that rdate only supports the old RFC-868 time protocol, so its a bit harder to find a server who supports this protocol. -- Thomas O. This area is designed to become quite warm during normal operation. |
| |||
| james <mentor@arisia.org> wrote: > I wrote an rc.ntpd with status|start|stop|restart with logging and a > logrotate configuration. That script seems really nice. However, I am missing one thing: A call to ntpdate before ntpd is started. If I remember right ntpd refuses to start if the clock is more than 3600 seconds wrong. 3600 seconds only means one hour. If you have a computer with a broken cmos battery or you don't have UTC time in cmos clock and switched between daylight saving the time might differ more than 1 hour from the correct time. With such a time difference ntpd might refuse to start when it is needed the most. A call to ntpdate fixes this. Another thing to consider for the script might be to write the right time back to the cmos clock with "hwclock -w". If the cmos battery is OK this will give a rather accurate time at the next reboot. Without the "hwclock -w" the time might be wrong at each reboot until ntp is started. regards Henrik -- The address in the header is only to prevent spam. My real address is: hc8(at)uthyres.com Examples of addresses which go to spammers: root@variousus.net root@localhost |
| ||||
| Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> wrote in news > james <mentor@arisia.org> wrote: >> I wrote an rc.ntpd with status|start|stop|restart with logging and a >> logrotate configuration. > > That script seems really nice. However, I am missing one thing: A call > to ntpdate before ntpd is started. If I remember right ntpd refuses to > start if the clock is more than 3600 seconds wrong. 3600 seconds only > means one hour. I believe that you're right, but since my 433MHz Celeron time server has a good CMOS battery I haven't run across that problem yet. ;v) > Another thing to consider for the script might be to write the right > time back to the cmos clock with "hwclock -w". If the cmos battery is > OK this will give a rather accurate time at the next reboot. Without > the "hwclock -w" the time might be wrong at each reboot until ntp is > started. I think that my slackware version, version 10.0, handles that as part of the reboot/poweroff scripts. Maybe??? Anyway, thanks for the kind words. You are, of course, free to take/alter/re-publish the script as you wish. After I wrote that I started checking out more of the ntp package. There're a lot of other scripts in the package but they don't seem to work, as if they were written for a different version or distribution. I'm a little vague as that was a while ago now. |