1

This question was asked a lot in the past, but since ntpdate has been depricated, I'm having issues finding answers that work via ntp.conf. I've read a lot about setting a cron job for ntpdate is a bad idea now, so that answer won't work for me.

My server time (ubuntu 14.04 server) is drifting quite a bit, and I have a time server I'm using locally.

In my ntp.conf file I have my server listed like this

server 192.168.1.25 iburst burst minpoll 4 maxpoll 4

Despite these settings the watchdog timer run from

ntpdc -c loopinfo 

never resets and my drift keeps getting larger and larger.

To sum up my question, why isn't my /etc/ntp.conf file actually doing anything? More info: This is a Ubuntu 14.04 VM running on Hypver-V and is used as a Redmine server.

output of ntpq -p (not sure if typo, but no q option)

    remote           refid        st  t  when  poll  reach   delay   offset   jitter
   ===================================================================================
    myother.server 192.168.1.26    2  u   2    16    377     0.607   0.808     4691.46 
  • Are you able to set the time using ntpdate? – muru Jul 17 '14 at 18:22
  • Please add the output of ntpq -qp. – Reinstate Monica - M. Schröder Jul 18 '14 at 14:10
  • When I run ntpdate -u 192.168.1.25 I get the offset, and I'm assuming it also sets the time correctly. However, when I try to set ntpdate up in cron, it won't run. I'm not sure why, but since it is depricated, I didn't really worry about it. – trueCamelType Jul 18 '14 at 18:36
  • The requested output is shown above. – trueCamelType Jul 21 '14 at 20:09
  • With that output it is clear it thinks your source is Stratum-2, which is very good. And surprising, since you are not using time.nist.gov. Here is a reference Q&A that talks to this subject, suggesting you reference the master clock directly: askubuntu.com/questions/972799/… – SDsolar Nov 4 '17 at 8:09
0

After a bit of searching, the problem was obvious. It wasn't actually a problem with the machine itself, but with hyper-v. Apparently Hyper-v is a terrible choice for linux virtualization. It's a known problem that doesn't appear to have a fix other than just running something on the guest Linux machine.

I use this cron job

0 * * * * /usr/sbin/ntpdate 192.168.1.x

obviously the 192.168.1.x represents the ntp server. This fixed my time every hour. The documentation on how hyper-v has a hard time keeping time with linux guests is noted in several places such as here, and here.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.