VMware Cloud Community
jcurtiss7492
Contributor
Contributor
Jump to solution

ESXi 6.7 - NTP time drifting on one host

I have (2) ESXi 6.7 hosts setup to use the same NTP server (time.nist.gov).  One host keeps the time fine, even though sometimes the logs will display a warning that it has lost connection to the NTP server.  The other host loses time gradually.  When I noticed the problem, I simply restarted the NTP Daemon and the time was corrected and in sync with the first host.

One week later, there is a 5 second difference between the 2 hosts.  The NTP service is configured exactly the same on both hosts.  Any suggestions on what I should try next?  I figured I'd try the forums before contacting support.

0 Kudos
1 Solution

Accepted Solutions
NathanosBlightc
Commander
Commander
Jump to solution

NTP is a backward compatible service... however, you can specify its version on the ntp.conf file (check one of my attached pictures on the above link in my previous reply to know how to do it). So please check the status of the operation via "ntpq -p" command and if it's possible, try to force it to use IPv4 instead of IPv6 and then tell me the result ...

Please mark my comment as the Correct Answer if this solution resolved your problem

View solution in original post

0 Kudos
8 Replies
NathanosBlightc
Commander
Commander
Jump to solution

If you ensure about the ESXi hosts have the same version and build number, and both of them have similar hardware platform, and NTP configuration is same, please check the NTP synchronization network traffics to find-out what's going on both of your hosts, via shell/SSH access:

tcpdum-uw -i vmk0 host x.y.z.w and port 123

Also, you can check the time settings via cat /etc/ntp.conf file. I explain about how to troubleshoot the ESXi NTP settings, in the following link:

Time differentiate between ESXi host & NTP Server

Please mark my comment as the Correct Answer if this solution resolved your problem
0 Kudos
berndweyand
Expert
Expert
Jump to solution

check with "ntpq -p" if time is synched periodically

jcurtiss7492
Contributor
Contributor
Jump to solution

Hello Amin, thanks for your reply and info.

I went thru the article and noticed a few things.  Host 2 that works, I can see the ntp request go out and the reply come back.  However, Host 1, that doesn't work, shows the reply go out, but no response come back in, like in these 2 screenshots.  I checked the firewall settings on both hosts and they are identical.

NTP-host1.png

I actually did get 1 reply as you can see, that was when I manually restarted the NTP service.  I feel like this should be telling me something but I'm not getting it.  What am I missing?  For reference here is a screenshot my other host that is working.

NTP-host2-working.png

I can see that the NTP version on the client is v4 and the version on the nist servers are v3 but since the time is working correctly on the one host, I feel this isn't my cause.

Thanks!

0 Kudos
NathanosBlightc
Commander
Commander
Jump to solution

NTP is a backward compatible service... however, you can specify its version on the ntp.conf file (check one of my attached pictures on the above link in my previous reply to know how to do it). So please check the status of the operation via "ntpq -p" command and if it's possible, try to force it to use IPv4 instead of IPv6 and then tell me the result ...

Please mark my comment as the Correct Answer if this solution resolved your problem
0 Kudos
jcurtiss7492
Contributor
Contributor
Jump to solution

Does anyone know of a way to force the NTP service to use IPv4 instead of IPv6?  I've gone through all of the VMware NTP articles and couldn't find anything on this.

0 Kudos
berndweyand
Expert
Expert
Jump to solution

try it with "server -4 time.nist.gov" in the ntp.conf

jcurtiss7492
Contributor
Contributor
Jump to solution

Thanks Amin!  Issue cleared up after forcing it to use IPv4 instead of IPv6.

0 Kudos
jcurtiss7492
Contributor
Contributor
Jump to solution

Thanks for the help bewe, that "-4" worked to force IPv4!

0 Kudos