I have 2 Sun T5220 machines in the lab, which synchronize time with a Sun 
Netra240 machine every 15 minutes. However, large clock skews are occasionally 
found on these two machines. Then I put a script on crontab which runs every 5 
minutes to monitor the time difference with the time server. The content of 
this script is as follows:
crontab -d TimeServer |tail -1 >> /var/tmp/timediff.txt
When the clock skew is found, I can see from timediff.txt that one of the 
machine has executed the script, while the other missed one time:
T5220#1:
23 Oct 11:00:00 ntpdate[8338]: ntpdate 3-5.93e+sun 03/06/05 23:16:45 (1.4)
23 Oct 11:05:00 ntpdate[9361]: adjust time server 10.170.7.84 offset -0.001842 
sec
23 Oct 11:10:00 ntpdate[10494]: adjust time server 10.170.7.84 offset -0.068542 
sec
------------------------------------------------------------------------------------------------
T5220#2:
23 Oct 10:55:00 ntpdate[4872]: adjust time server 10.170.7.84 offset -0.078087 
sec
23 Oct 11:00:00 ntpdate[5878]: ntpdate 3-5.93e+sun 03/06/05 23:16:45 (1.4)
23 Oct 11:10:00 ntpdate[1816]: adjust time server 10.170.7.84 offset -0.045485 
sec
23 Oct 11:15:00 ntpdate[3023]: ntpdate 3-5.93e+sun 03/06/05 23:16:45 (1.4)

You can see that on T5220#2, the script was not executed at 11:05:00, I suspect 
that the clock on T5220#2 was set to a time after 11:05:00 by ntpdate so it 
missed the one at 11:05:00, and the next time it sync time with the server, the 
clock was corrected, if this kind of problem really happens, it lead to 
troubles on database replication. Can anyone explain this issue and hopefully 
tell me how to fix it?
Thanks!
--
This message posted from opensolaris.org
_______________________________________________
sysadmin-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/sysadmin-discuss

Reply via email to