Option 50 in failover mode

[Bob Proulx]

Abu Abdulla alhanbali wrote:
> Any hints please. it's difficult for me to find the information from
> the RFCs or the list.
> 
> Abu Abdulla alhanbali wrote:
> > i would appreciate if someone can identify the cases where the server
> > ignores option 50 "dhcp-requested-address".
> >
> > i have a failover mode in which the client needs to sustain the same
> > IP (if rebooted) as far it is not expired. in some cases the "secs"
> > field exceeds the max load balance resulting the other server to
> > response with different IP. we implement option 50 in the client and
> > now the second server starts to give us the same IP:
> >
> > - is it guaranteed that this will be the behavior all the time?
> > - is this assignment because the two servers are synced and the second
> > server is just informing the user that your previous IP is still
> > active. what will happens if this is done before syncing? what is the
> > period between syncing?
> > - will a server ignore option 50 in case the ip is in the backup pool.

The dhcp failover mode is described in the dhcpd.conf manual in the
"DHCP FAILOVER" mode section.  That section answers your questions in
detail.

When two dhcpd servers are configured in failover mode then it is
critical that you have enough IP address space in your pool that
either single server's pool is large enough for the entire network.
This can sneak up on people since if both are operating you
effectively have twice as much address space available.  You could use
more than available in a single server and never know it.  But if one
system is offline then the remaining system will need to handle the
entire network itself in its half of the pool.  Therefore each server
needs to have a pool size sufficiently large to handle the entire
network itself.  If you do then the answer to your questions is yes.
If you do not then the answer to your questions is no.

At this point several people will jump in talking about putting the
single remaining server (after the other failed) into partner-down
mode in order to move allocatable ip space from the down server to the
remaining server.  That is all well and good but if you are available
to do that then you would also be available simply to repair the
failed server and get it back online.  Personally I am only interested
in automatic, hands-off high-availability and not manual, hands-on
high-availability.

Bob