Maximum Time difference between DHCP primary and DHCP failover Server

[Simon Hobson]

kraishak <kraishak.edu at gmail.com> wrote:

> Actually I want add validation script so that for the given two
> server address it can calculate the time difference and give me the results
> for proceeding to add failover peer on not.

Fair enough - seems a reasonable check to do. But I have a suggestion for an alternative approach ...

> Can you also please comment or share your experience if you have any, about
> what could be the maximum network latency between the primary and failover
> DHCP servers that does not impact the server performance 

I've not actually used failover, it's just not been appropriate to my needs. But given the answer you've been given of "anything within 60s is OK", I don't think you should have a problem.

I would tackle the issue from a different direction.
Put measures in place so that you can assume all your servers are time-synced, and have monitoring in place so you can see if any of them aren't. That way, you can assume DHCP failover will work (or at least, time won't be a reason for it to fail) as long as you don't have any alarms in place.

Obviously NTP is good for keeping everything in sync, that's a problem solved decades ago !
At my last job, I used Nagios to monitor all sorts of stuff. One of the things it can monitor is NTP status - raising alerts if the monitored server isn't synced with a source, or (from memory) if it's time differs from a configured source by more than a settable amount. Monitoring was something I made a big thing as I'd gone into an environment where even basics like time sync weren't being done, and monitoring was of the "phone rings because a customer sees a problem" style !

So with NTP working you don't have a time problem. With a monitoring system in place, you'll know if there's a problem with that working.
If your monitoring system has a green status page, time differences aren't a consideration for your DHCP deployment.


Simon