~16, 000 backup leases added to dhcpd.leases when implementing failover

[Shawn Routhier]

On Sep 17, 2014, at 10:35 AM, Mark Wagner wrote:

> When converting a single dhcp server to a failover configuration a large number backup leases were added to dhcpd.leases.
> 
> Here is the config:
> 
> subnet 10.63.0.0 netmask 255.255.0.0 {
>   filename "/pxelinux.0";
>   next-server 10.60.0.107;
>   option subnet-mask 255.255.0.0;
>   option routers 10.63.0.1;
>   pool {
>     failover peer "dhcp-failover";
>     range 10.63.128.1 10.63.255.254;
>   }
> }
> 
> The leases start around 10.63.191.100 and go to 10.63.255.255. An example lease is
> 
>  lease 10.63.191.171 {
>   starts 1 2014/09/15 23:06:08;
>   tstp 1 2014/09/15 23:06:08;
>   tsfp 1 2014/09/15 23:06:08;
>   atsfp 1 2014/09/15 23:06:08;
>   binding state backup;
> }
> 
> Is this normal? It didn't seem to happen for other subnets.

A binding state of backup means that they are available for the secondary to hand out.

When you have a pool being served by both peers of a failover relationship the currently
unused leases are split between the two to provide better responses.  The backup tag
is how the servers indicate this in the lease file.  

You should be seeing some leases in the backup state for all pools that are declared to
be part of a failover relationship.

> 
> How can I safely remove these leases?

No

> 
> 
> -- 
> Mark Wagner <mark at lanfear.net>
> _______________________________________________
> dhcp-users mailing list
> dhcp-users at lists.isc.org
> https://lists.isc.org/mailman/listinfo/dhcp-users

Shawn Routhier
ISC
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/dhcp-users/attachments/20140917/d06b49f4/attachment.html>