There are a couple of scenarios when this can occur -
- If the DHCP server is rebooted, the DHCP IP allocation table also gets cleared (This can be seen using the "show vlan <name> dhcp-address-allocation" command. This means that the switch is currently not "aware" of the IP addresses that has been offered. So, when the next discover packet comes in to the server, the server may try to allocate an IP that has already been allocated earlier (and whose lease timer has not expired). The server will then continue to try allocating other IPs until a "free" IP is allocated to the client.
- This can also occur when the "clear vlan <name> dhcp-address-allocation" command is issued which will again, clear the dhcp allocation table and the behaviour is the same as mentioned above.