Lease binding state update triggers?

[Tom Griffin]

This is happening on multiple networks now and far too frequently. In 
order to stabilise the system, I have simply disabled failover for the 
subnets which are affected, this means running in a non-resilient 
configuration which obviously is a bad thing. I has happened on subnets 
with a lease time of 24 hours and 7 days, so this doesn't seem to have 
any bearing on the frequency for me.

Is there any debugging I can perform to try and find the cause of this 
problem?

Luis Fernando Lacayo wrote:
> Good Morning all,
>
> I had this happened to me on a few of my networks, What I did was to 
> reduce the lease time, and when all the leases which had a longer 
> lease expired, I saw it happened a lot less.  It still happens but not 
> as frequent.
>
>
> On Wed, 2008-08-06 at 12:55 +0100, Tom Griffin wrote:
>> Hello,
>>
>> We have a problem where users cannot get a lease on a subnet configured 
>> with failover on some occasions. The error in the DHCPD log is "Peer 
>> holds all free leases" which is shown by both servers.
>>
>> Looking into the leases file, I can see that there are no leases in the 
>> "free" state, they are in other states like "expired". All these leases 
>> show "next binding state free" and the "ends" time has already passed.
>>
>> When does DHCPD process the lease file to determine which leases should 
>> have their binding state changed and is there any reason why this would 
>> not happen for a given subnet?
>>
>> Could this be caused by the fact that our scripts could potentially 
>> restart both dhcpd servers every 5 mins in order to add new hosts? I 
>> realise this is something omshell can do, but its documentation is 
>> lacking so I haven't gone down that route. I am more willing to give it 
>> a try if it will stop future issues like this.
>>
>> Many thanks,
>> Tom.
>>
>>     
> -- 
> Luis Fernando Lacayo
> Chicago Public Schools
> Senior Unix Administrator
> ITS/ UNIX Infrastructure
> Office: 773-553-3835
> Cell: 773-203-4493
>