Wireless Access

Ok, we only use the on-controller dhcp server for the guest portion of our wireless deployment, which in most cases is very small.

We recently put in a new controller at a new site, and set up the guest ssid and catpive portal, etc, but it stopped working earlier today.  This is what I found-

lease 172.20.0.253 {
  starts Thu Nov 15 20:33:13 2012
  ends Thu Nov 15 21:03:13 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:14 2012
  ends Thu Nov 15 21:03:14 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:15 2012
  ends Thu Nov 15 21:03:15 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:16 2012
  ends Thu Nov 15 21:03:16 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:17 2012
  ends Thu Nov 15 21:03:17 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:19 2012
  ends Thu Nov 15 21:03:19 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:20 2012
  ends Thu Nov 15 21:03:20 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:21 2012
  ends Thu Nov 15 21:03:21 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:22 2012
  ends Thu Nov 15 21:03:22 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:23 2012
  ends Thu Nov 15 21:03:23 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:24 2012
  ends Thu Nov 15 21:03:24 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:25 2012
  ends Thu Nov 15 21:03:25 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:27 2012
  ends Thu Nov 15 21:03:27 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}
lease 172.20.0.253 {
  starts Thu Nov 15 20:33:28 2012
  ends Thu Nov 15 21:03:28 2012
  binding state active;
  next binding state free;
  hardware ethernet 38:aa:3c:f6:28:e1;
  uid "\0018\252<\366(\341";
}

 This actually would go on and on and on, several hundred entries of the same lease to the same mac.  Also I should point out that the clients never get an address.

I restarted the dhcpd process, didn't fix it, restarted the controller, no joy, and even deleted the dhcp server settings, restarted the controller, and reconfigured the dhcp server and it didn't do anything different.

Anyone seen this before?  It's not a huge issue at the moment but I will opening a TAC case tomorrow unless someone has a fix for it.

Hummm...

What version are you running?

Also, can you paste/post up your DHCP and IP interface config portions please?

Also if you see the output the lease start time increment to 1 sec for every entry. can you also check if the client is renewing the IP continuously. it could be because of a virus on the client. you can verify this with a wireshark capture on the client machine. or try enabline debug on the dhcp. 

#conf t logging level debuging network process dhcpd

and check the "show log network 30" output.

Well this is a really interesting issue... i have seen this but i mean not hundreds  of times... i have seen this but maybe just like repeating 10 times... and well actually didnt give me any issue but still ill be looking forward for the resolution... i would like to know what code version you got? it is an old 3.x.x.x version?

Other thing which might help you is, assuming it's a normal person with a malfunctioning client device, handy to know it's a Samsung one (OUI check). Or at least has a Samsung NIC.

Assuming the client is wireless rather than wired, why not look at him on the client list and blacklist him to test?

Knowing who it is will help a lot.

Sorry for the lack of responses, was busy working on higher priority items.

It's actually several clients, not just one MAC that has the issue or the repeated leases.

It goes on and on until eventually the dhcp server just freezes up and no clients can get on anymore.  I clear the bindings and restart the process to get it going again but that is definitely not a good long term solution.  Luckily most people on this particular SSID are just there for 30 minutes or less.

What code version you got? maybe it could be a bug...

6.1.3.4

I have a customer running 5.0.4.8 on a 3200 that is experiencing the same issue with a client, though it only seems to be happening to a single client.  Client is a Apple device.

Anyone have a fix for this? 

Weird,  the clients attempts dried up, now i only see 2 leases in the controllers dhcp server for them.

Maybe its just a client issue?