NTP Drift, How To Fix?
So I have a mixed environment of RHEL 5/6 servers and on some of them, there is ntp drift. The time drift varies for all of them.
These are all VMs. All of them are patched up and I have iptables rules to allow for 123 on input/output.
For one of them, when I run various ntp commands, such as:
ntpstat
Unable to talk to NTP daemon. Is it running?
./ntpd status
ntpd dead but pid file exist
ntpdate -u (ntp server address here)
18 Nov 09:4700 ntpdate (11967): step time server 143.83.220.1 offset --197.500050 sec
The final command forced it to work, however I will check it later this week and there will be drift.
Also what is the general rule for having the hardware clock match the system time?
thanks
Responses
Hello
You did not say what VM system you are using but I think the principle is the same. The VM system simulates the hardware clock, but it is not as stable as you might expect as the hypervisor updates the simulated hardware clock in a non-deterministic way. The first thing would be to check you have at least three NTP servers entered in the config file. Searching for more stable clocks might also be required.
Using two NTP sources is considered to be the worst choice as the NTP application, ntpd in this case, cannot decided which is more stable. Have you configured the host system to act as an NTP server to the guests?
If you search the Knowledgebase you will find many articles and solutions that may help. I remember this previous thread About behavior of ntp which has links to various parts of the guides which might help.
Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.
