Michael wrote:
We tested the patch you suggested with no success. We are seeing timestamps straying into the 100's of seconds of difference, which does not reflect the perceivable drift that shows on our systems clock (which is negligible, sub 1 second). Currently we run against two stratum 2 servers that get their time from two stratum 1 servers, and per Stan's suggestions earlier we are rebuilding the stratum 2 machines on bare metal hardware (not in a virtual machine) to be sure the clocks will be super stable. I guess what I am asking is if you have ever seen an issue similar to this and if NTP actually played a role. I have spent some time reviewing the mailing list archives and have not found a definitive answer from other Dovecot user's experiences.
(posting from my work account...)
Barring strange hardware or OS behavior, NTP is generally really good at what it does.
There are known problems with some hardware and some OSes (and virtual environments in particular).
See http://support.ntp.org/bin/view/Support/TroubleshootingNTP for more information.
And just to try the direct approach, if correct time and NTP are important to you, please join the NTP Forum. We need the support.
-- Harlan Stenn <stenn@ntp.org> http://ntpforum.isc.org - be a member!