I have a problem with CCr2216, ROS 1.18.2, regarding DHCP server and IP addresses that are not released by the DHCP server.
At times when all users drop (due to an blimp on the ENNI for example), We have pool full errors for about 10-20 minuntes as the DHCP still thinks the IP’s are assigned to users.
I have several /24 IP pools, eg 150.119.4.0/24, 150.119.5.0/24, 150.119.6.0/24, 150.119.7.0/24 etc
When clients on any pool tried to release their IP address / renew, broadcasts are sent to 150.119.4.1 and client CPE gets a DHCP Server not found message.
Should those broadcasts not go to 150.119.4.1, 150.119.5.1, 150.119.5.1, 150.119.6.1, 150.119.7.1 etc based on the pool from which the users IP address was assigned?
Hi,
The DHCP will continue to reserve an address for a particular client until the lease period has expired. Any disconnection between the client and the server does not cancel a reservation so that is why you are experiencing the “IPs are assigned to users”.
When a client does a renew it broadcasts to all DHCP servers and then chooses one to accept. The exception to this is if you have DHCP relay functionality implemented. All this happens at layer 2 rather than the IP address layer because the client doesn’t yet have an IP address assigned to it so can’t talk to another IP address.
If the same DHCP server is selected by the client, the server should recognise the MAC address and reissue the same IP address.
I am not sure why you are not getting a server not found error unless something is overloaded.
In your implementation, how are you assigning an address pool to a requesting device?
Peter.