I have a 3750 that has a few different DHCP pools setup on it, 1 of the pools is for a guest wireless vlan that seems to be giving us issues. Wifi is running on Meraki APs, the APs are configured as trunk with all vlans allowed. Our internal wifi is bridged to vlan 1, those clients get DHCP leases with no issues but clients on Guest Wifi often can't connect and get "unable to obtain IP address" message on their phones. We use the same config at approximately 20 other sites with no issues.
Meraki has went though the config on their side and verified no issues, our networks are all cloned as well so I didn't assume any config issue in the Meraki cloud was the problem. The switches show DHCP bindings to the guest wifi vlan but we still continue to have widespread issues out of this particular DHCP pool.
--> Our internal wifi is bridged to vlan 1, those clients get DHCP leases with no issues but clients on Guest Wifi often can't connect and get "unable to obtain IP address" message on their phones.
Can you post a schematic drawing of your topology, showing what 'internal wifi' and 'bridged' mean in this context, that is, a drawing that shows the physical and logical topology ?
Thanks for posting the switch configurations. The main thing that I notice on the 3750 is that you have configured 2 DHCP pools using exactly the same address range and other parameters. Remove one of those pools. (done know if it makes any difference which one you remove).
On SW2 am I correct in understanding that the interface where Meraki is connected is G1/0/13? The output of show trunk does indicate that G1/0/13 does have 999 as an active vlan. That looks encouraging. On SW2 I see several mentions of network 10.10.10 (including an OSPF network statement) but I do not see any interfaces with that network. What can you tell us about 10.1.10?
Ok, I have removed OSPF and everything related to 10.10.10. Also the 2 DHCP pools never really existed, when I was scrubbing the config to post I saw the second "guestwifi" pool and just the name existed, no networks. I thought I had removed the guest pool when I was troubleshooting last week so I filled it in on the text doc, not seeing the "Guest" pool above it. As of right now this is the only DHCP pool for vlan 999:
ip dhcp pool Guest
network 10.99.83.0 255.255.255.0
dns-server 22.214.171.124 126.96.36.199
I have set Meraki back to use vlan tag 999 so I'll see if DHCP errors increase again.
So some devices are successful in getting IP and more are failing? Here are some things we might look into:
- does the Meraki have any information about the failures? any error codes? any information about the devices that are failing?
- is it possible that any of the AP are not using vlan 999?
- perhaps debug for DHCP on the switch doing DHCP might shed some light?
- perhaps a fresh copy of the configs?
perhaps the output on both switches to the commands show cdp neighbor, show interface trunk
The Meraki's don't show anything in logs other than "disassociation due to DHCP failure" and Meraki support verified there were no issues on the Meraki configs.
Both APs are set as Trunks with all Vlans allowed
I had ran debugs but everything looked good, any suggestions on which debugs to run?
I've attached the configs again. The weird thing is we get plenty of leases in the switch, it's like the failure is is between the OR of the "DORA" process, the switch is receiving the offer and creating a lease but it's not getting back to the client.