Failover configuration file errors?

Jason Gerfen jason.gerfen at gmail.com
Thu Sep 25 16:08:43 UTC 2008


I am glad you mentioned this to me because I am incorporating some
failover options in web based interface to assist in management of the
dhcpd service and because I am very unfamiliar with the failover
options I will (now that you mentioned some of the values and their
effects) have to research these options further to perhaps place
limitations on the values that can be entered.

But to answer your question, no, that is not what I planned.

On Thu, Sep 25, 2008 at 10:02 AM, Alan Clegg <Alan_Clegg at isc.org> wrote:
> Jason Gerfen wrote:
>
>> failover peer "tyr.scl.utah.edu" {
>>      primary;
>>      address 155.97.15.2;
>>      port 519;
>>      peer address 155.97.15.1;
>>      peer port 520;
>>      max-response-delay 30;
>>      max-unpacked-updates 5;
>>      mclt 5;
>>      split 2;
>>      load balance max seconds 30;
>> }
>
> Since you posted the entire stanza....
>
> I'm quite curious as to your choice of split.  What was the rationale
> behind the choice of 2?  This will force almost (but not quite) all
> lease responses to be given by the second machine in the failover pair.
>
> Was this what you planned?
>
> Additionally, an mclt of 5 seconds is going to cause some MAJOR
> thrashing of your network when either a server goes down or they can't
> communicate for any reason the longest lease extension available to
> clients will be 5 seconds long...
>
> AlanC
>
>



-- 
Jason Gerfen
jason.gerfen at gmail.com

~ tomorrow ain't promised so we live for today


More information about the dhcp-users mailing list