Classes and failover

[Austin Gabel]

The classes should not have any affect on this issue.  Your failover is the
problem here.  Try adding the split line to the secondary failover
declaration.  I'm pretty sure this statement is required for both servers.




On Thu, Jun 26, 2008 at 9:47 AM, Rene Joergensen <rgj at fzr.dk> wrote:

> Hi there,
>
> We have a problem with dhcpd 3.1.1 running in failover.
>
> As i understand the implementation, when receiving a request, the two
> servers calculates who should reply to the request (configured via the
> "split" option).
>
> What we're seeing is that both servers reply on requests which never
> reaches the client, maybe because the relay in the Cisco/Alcatel
> equipment gets confused.
>
> But why do both of them reply?
>
> We use classes to limit the number of leases:
>
> class "class3" {
>        match if exists agent.circuit-id;
>        spawn with option agent.circuit-id;
>        lease limit 2;
> }
>
> And for circuits with fixed ip, there is a class defined:
>
> class "TN-222827" {
>        match if option agent.unknown-6 = "TN-222827";
>        lease limit 1;
> }
>
> And we then define a pool with the fixed IP, a pool with only one IP in
> the range, and deny them in the other pools.
>
> What we see in the logs:
>
> Jun 26 14:34:31 choke dhcpd: DHCPREQUEST for 194.19.142.187 from
> 00:07:e9:af:35:45 via 194.19.167.1
> Jun 26 14:34:31 choke dhcpd: DHCPACK on 194.19.142.187 to
> 00:07:e9:af:35:45 (eeyore) via 194.19.167.1
>
> Jun 26 14:34:31 serv01 dhcpd: DHCPREQUEST for 194.19.142.187 from
> 00:07:e9:af:35:45 (eeyore) via 194.19.167.1
> Jun 26 14:34:31 serv01 dhcpd: DHCPACK on 194.19.142.187 to
> 00:07:e9:af:35:45 (eeyore) via 194.19.167.1
>
> Normally it logs that is loadbalances, but not here.
>
> Our configuration:
>
> On the primary dhcpd:
>
> failover peer "dhcp-failover" {
>  primary;
>  address 194.19.194.18;
>  port 647;
>  peer address 194.19.194.29;
>  peer port 647;
>  max-response-delay 30;
>  max-unacked-updates 10;
>  mclt 600;
>  split 128;
>  load balance max seconds 3;
> }
>
> On the secondary dhcpd:
>
> failover peer "dhcp-failover" {
>  secondary;
>  address 194.19.194.29;
>  port 647;
>  peer address 194.19.194.18;
>  peer port 647;
>  max-response-delay 30;
>  max-unacked-updates 10;
>  load balance max seconds 3;
> }
>
> Any ideas why? Is it because of the classes?
>
> --
> -René
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/dhcp-users/attachments/20080626/f44a017c/attachment.html>