If you have noticed any of the following issues, they may be related to the NTP offset on the hosts that are reporting metrics through the Agent:
To check the NTP offset for a host, you can run the Agent status command, using the instructions appropriate for your OS, and look for the Clocks section:
Clocks ====== NTP offset: -0.0036 s System UTC time: 2015-02-12 22:10:49.524660
Any significant offset can have undesired effects, and it is recommended to keep hosts in sync with NTP to ensure this does not cause issues. To help with this, there’s an NTP Integration.
Instructions for syncing the system clock with NTP vary based on the operating system being used. Here are some helpful links for different operating systems.
To prevent NTP issues, you can leverage Datadog’s monitor for NTP offset to alert you when there is drift on a host.
Also, you can check over all the reporting hosts for one that has offset issues: use Datadog’s Check Summary page. Inspect the check
ntp.in_sync and see the list of the hosts that have NTP issues.
Also note that outgoing UDP traffic over the port 123 should be allowed so the Agent can confirm that the local server time is reasonably accurate according to the Datadog NTP servers.
Additional helpful documentation, links, and articles: