Thanks for posting the packet captures. I am afraid that they do not shed much light on the problem. They show that the machine in VLAN 6 is sending requests but gets no response and that the machine in the same VLAN as the server does get a response. But we pretty much knew this already.
I do find it interesting that the machine in the same VLAN as the server sends multiple requests. The first three requests do not get any response. It is only after the fourth request that the client gets a response from the server.
Can you tell us how the server is connected? Is it connected to a port on one of the switches that you have mentioned? Would it be possible to get a span port configured to monitor the traffic going to the server and responses from the server?
We have the main router 3750 and a couple of 4506 in different floors. The server is connected to one of the 4506.
I dont know how to do a span port, do you have a link to a document with more info ?
Thank you so much for all your help!!
Here is link on how to configure SPAN ports on the 4500 series switches:
The link given by Reza is a good one. You would want to set up span/monitor with the port where the server is connected as the source and some available port as the destination. Then you would connect a PC that can run Wireshark to the destination port and do the packet capture. We would want the capture to verify whether the DHCP request is sent to the server and what, if any, response is sent by the server.
Also post the config of the port that is connected to the DHCP server
Thank you all for your posts, sorry for my delay in replying ...
After 3.5 hrs in the phone with Cisco, the problem is resolved.
1) One machine was broadcasting arp messages to all the network creating a high CPU utilization (router).
2) After that machine was disabled the network started to be stable again, but DHCP problem wasn't resolved.
3)Then we turned off the VTP pruning, DHCP problem was resolved and machines were able to get IP by DHCP.
The machine that was creating the traffic is a UNIX box with dual NIC, the first NIC with static IP in vlan 13 and 2nd NIC with an IP in a non-routable vlan(192.168.x.x). Another UNIX box in the same vlan (13) was using the same non-routable IP (192.168.x.x) causing the conflict with IP address and letting one machine sent out all the arp requests.
Thanks for your valuable time !
Thank you for posting back to the forum indicating that the problem was solved and how you fixed it. VTP pruning was certainly not one of the things that I had thought about as a possible cause of this and it is interesting that this turned out to be the problem. Now that you have it working perhaps you want to mark this question as resolved?