silently neglected requests fixed by restarting dhcpd

Glenn Satchell Glenn.Satchell at uniq.com.au
Tue May 9 17:16:55 UTC 2006


>From: "Gordon A. Lang" <glang at goalex.com>
>To: <dhcp-server at isc.org>
>Subject: silently neglected requests fixed by restarting dhcpd
>Date: Tue, 9 May 2006 10:16:42 -0400
>
>I'm looking for some help.
>We started using dhcpd 3.0.3 on March 28, 2006 for 15 out of our 45 VLAN's and 
it worked fine for a few weeks.  The other 30 VLAN's continued to be serviced by 
our Microsoft DHCP server.
>
>Then we added the rest of our VLAN's, and it worked fine for a while until the 
morning rush at which time there were too many problems to figure out so we 
backed out our change for most of the VLAN's.  This didn't help though because 
the problems were occurring on the VLAN's that had previously been working.  
Shutting down the failover partner, and restarting the primary server finally 
fixed the problem.
>
>Everything seemed fine for the last couple of weeks with the dhcpd server 
handling about 25 VLAN's.
>
>Today I tried once again to migrated the remaining VLAN's.  Again, everything 
seemed fine initially, but then problems began to occur.
>
>Renewals would work, but initial requests (forwarded by the routers via 
helper-address) were failing on some VLAN's.  The interesting thing is that most 
of the failures were occurring on the original VLAN's that have been working 
since March 28.
>
>The problems this time were different than last time in some important ways, 
but the problems persisted until I did a stop and restart of the dhcpd daemon.  
Since then everything seems fine again.
>
>We have two routers using HSRP and all VLAN's have helper-address 
configuraitons.
>
>Any ideas?

Got any log files showing the error messages? Posting a section from
your dhcpd.conf (we don't really need to see 45 subnet declarations)
and a brief description of your platform and network layout might help
us too.

Without specific errors it could be anything. Although 45 VLANs doesn't
sound that it would be an overload - I seem to remember users who had
several 10s of thousands of hosts running from one DHCP server.

regards,
-glenn



More information about the dhcp-users mailing list