vMotion makes NTP trigger, how to prevent?

Troubleshooting
1

CMK version: 2.3.0p19
OS version: RHEL 8.10

Issue: Problem is that CheckMK has a “attempts” set to 10, but still triggers NTP Time to be completely out of sync. This was suppose to prevent from triggering the notification, but it feels like 10 attempts, is like 10 minutes until triggering. Could someone perhaps explain how we can configure each attempt time, so that we have enough time before triggering. Right now we are unsure how much time each “attempt” uses up, before hitting the 10 attempts.

2

Each check attempt is 60 seconds in between as this is the usual check interval of the Check_MK service that fetches the agent data.

We have set the maximum number of check attempts for the “NTP Time” service to 60 for exactly that reason.

3

Could be a feature request ?
Setting the retry time from 60 seconds to something bigger ?
Since this could be done in Zabbix and Nagios as well.

4

No, as all checks get their data from the agent in the same interval.

5

Alright, thanks for your input :slight_smile:

6

Maybe you could also try to address this issue on the VM side:
After a vMotion event, the VM’s time is synchronized with the (new) host — if the option “Synchronize time with host” is enabled.
If your ESXi hosts are not perfectly in sync, this could be the reason for your NTP warning.

You might want to check out this KB article:
https://knowledge.broadcom.com/external/article?legacyId=1189

And afterwards, take a look at this blog post:

TL;DR:

  • Check NTP synchronization across all your ESXi hosts
  • Consider disabling periodic time synchronization with the host
7

In such cases I use the rule “Delay service notifications” to suppress the notification if the service is not critical for longer than x minutes.

8

This is not recommended any more by VMWare for several years now. The VM should always run an NTP service.