We sometimes have the problem that new servers have the wrong time in the bios, so there time can be off by a month.
When you suspend a VM in VMware and then unsuspend it, the time will be off too. Because NTP does not sync after a maximum offset, I am considering using tinker panic 0 in the /etc/ntp.conf.
What is the reason that there is a default maximum offset of 1000 seconds that causes NTP to stop syncing time? We are using Puppet to set up NTP, I am considering to make it set tinker panic 0 in the ntp.conf, so NTP will sync anyway. What are the disadvantages of doing this?
Best Answer
The cause for not syncing against a server whose time is so different is documented here:
In my current NTP configuration, also controlled by
puppet
, I force the synchronization with the server, both in thentp.conf
file, usingtinker panic
, and in the daemon settings (/etc/sysconfig/ntpd
), as described in thentpd(8)
manpage:I do this because I can trust the NTP server I'm connecting to.
The relevant portion of the module that applies to the clients is as follows:
And the contents of the referenced files are:
and:
The
hiera
part is missing here, but you get the idea.