DHCP Conflict After Power Outage

[xZamalek]

Hello Everyone,

I have this Scenario and i want to understand how it works in the background if someone faced it before.

-There is a DHCP Pool configured with a leased time of 10 days on the CE.

-APs and Switches are taking IP from this DHCP this is the standard solution.

-When both the Router and Switches come back up after a reboot , many dhcp conflict occurs and only few number of devices assigned an IP.

-More strange is , there is a conflict addresses for IP addresses that is not assigned or used by any device on the network.

-I can see a device obtained already an IP and this IP is not found int he DHCP binding table on the router , instead it's found on the dhcp conflict table.

My Question is , does the Router and Switches remember their lease database after a reboot ? also who is responsible to assign the same IP to the Switch? will the Switch request it's old IP from the Router or the Router will have a lease database which will assign the same IP to the same Host ? or both the Switch and Router have this information? it's kinda strange that the DHCP Server on the Router already assigned an IP to the Switch and then detect it's a conflict and how there is a conflict addresses for IP addresses that is not assigned or used by any device on the network?

[marce1000]

                 >.... does the Router and Switches remember their lease database after a reboot ?
 - In general and also in this specific examples which proofs my point ; don't use routers and switches for DHCP ; it's better to separate DHCP from layer 2 and 3 services , and look into appliances , which many vendors offer with a clustering capability offering extra resilience for the situation(s) that you describe , 

 M.

[xZamalek]

It's not an option now to configure a centralized DHCP Server.

[MHM Cisco World]

clear ip dhcp conflict

Try clear ip dhcp conflict then check dhcp

MHM

[xZamalek]

Thank you , this will solve the problem partially , It doesn't work for long as the power goes up/down frequently , i also saw that we can configure a script to auto clear the binding , i just want to understand what is happening on the background until i find a permanent fix to this issue.

[MHM Cisco World]

show ip dhcp pool
show ip dhcp conflict 

check both when issue appear again, it give you some hint what happened with DHCP POOL.
MHM

[xZamalek]

Thank you , what happened is that the Router found DHCP Conflicts , added this conflicts to the excluded addresses , then no addresses were available to be assigned to the hosts.

[MHM Cisco World]

when the power ON/OFF 

the client dont send release to DHCP server or re-binding 
the DHCP keep client IP in binding table wait it lease timeout and client ask new IP. 
OR reverse client keep IP and DHCP Pool is clear it statistic 
the best if this happened again then you can use EEM, even power UP/Down action clear ip binding 
MHM

[xZamalek]

"the DHCP keep client IP in binding table wait it lease timeout and client ask new IP."

Will this be the case even if the DHCP Server is rebooted also ? because in this case the client and Server are both a Switch and a Router respectively.

[Richard Burts]

We do not know the details of how the DHCP on the CE was configured and that might make some difference in how to answer the question. But in general when DHCP is configured on router or switch and that device goes down, then when it comes back into service it has no knowledge of previously assigned addresses and starts assigning addresses from the beginning. But some devices in the network had assigned addresses, the device never went down so it retains its IP address and conflicts result when another device is assigned that IP. In that context the suggestion from M to not configure DHCP on router or switch makes sense.

I am not entirely clear about this statement "I can see a device obtained already an IP and this IP is not found int he DHCP binding table on the router , instead it's found on the dhcp conflict table." My guess is that what is going on is that a device had an assigned IP, after the router rebooted it assigned that IP to another device. When the conflict was noted the server removes the address from binding table (it is not valid there) and puts the address in the conflict table.

I do not understand this statement and would appreciate some additional explanation "More strange is , there is a conflict addresses for IP addresses that is not assigned or used by any device on the network."

[xZamalek]

Thank you , i will make it more clear for the 2 points.

-Both Router and Switches went down at the same time so the device (Switch or AP) should obtain an IP again from the DHCP on the Router which was Also rebooted , so the Router itself assigned an IP to the Switch and then the Router detected this IP is a conflict , how can the Router re-assign the same IP to another Device and detect it as a conflict if the Router leased this address?

-Regarding the 2nd Point , there are 5 devices on the site which require 5 IP Addresses , the DHCP Server detected there is a conflict address for example (192.168.1.10) which i can see it's not used or leased by any Device..So how the router detect there is a conflict for an Address which is not in use in the network ?