'peer holds all free leases' query

Nick Garfield Nicholas.Garfield at cern.ch
Mon Feb 26 14:16:51 UTC 2007


The mclt must be less than the lease time for the lazy updates to be
meaningful.  Take a look at my previous posts to understand how this
should be configured and the affect it has on the DHCP state machine.

> -----Original Message-----
> From: dhcp-users-bounce at isc.org 
> [mailto:dhcp-users-bounce at isc.org] On Behalf Of Pete Clarke
> Sent: Monday, February 26, 2007 12:41 PM
> To: dhcp-users at isc.org
> Subject: Re: 'peer holds all free leases' query
> 
> mclt 1800
> default lease time is 30mins
> 
> Cheers
> 
> ----- Original Message ----- 
> From: "Nick Garfield" <Nicholas.Garfield at cern.ch>
> To: <dhcp-users at isc.org>
> Sent: Monday, February 26, 2007 10:44 AM
> Subject: RE: 'peer holds all free leases' query
> 
> 
> > What values have you set for MCLT and default-lease-time? 
> > 
> >> -----Original Message-----
> >> From: dhcp-users-bounce at isc.org 
> >> [mailto:dhcp-users-bounce at isc.org] On Behalf Of Pete Clarke
> >> Sent: Saturday, February 24, 2007 11:34 PM
> >> To: dhcp-users at isc.org
> >> Subject: Re: 'peer holds all free leases' query
> >> 
> >> > I am one of those, who regularly get 'peer holds all free leases'
> >> > message in dhcp log files in our failover environment. 
> Occasionally,
> >> > I have to stop the daemon and purge lease database on one of the
> >> > server, when the problem gets particularly worse. We do 
> use NTP to
> >> > sync time. I am using V3.0.4. Is there a feature in version after
> >> > 3.0.4 that can help to alleviate this issue? I do see a 
> mention of
> >> > 'Failover servers try harder to retransmit binding updates upon
> >> > entering NORMAL state' in the release notes of 3.0.5rc2. Is 
> >> it related?
> >> 
> >> I also get a lot of these.
> >> I have a fairly complex setup, lots of pools (~9,000) in a 
> >> failover setup. I 
> >> get a lot of the "peer holds all free leases" messages.
> >> Server 2 doesn't seem to give any addresses out for certain pools.
> >> 
> >> Our setup is running 3.0.5 on Solaris. Would this just be 
> >> "one of those 
> >> things" - we can't simply down one box to purge the leases - 
> >> this is a 
> >> highly critical environment.
> >> 
> >> Any thoughts?
> >> 
> >> 
> >> Cheers 
> >> 
> >> 
> >> 
> > 
> >
> 
> 


More information about the dhcp-users mailing list