A customer of mine is having quite a few disconnection errors on the terminals, so he sent me the syslog from RFS7000 which is adopting all the AP650. I noticed there were a lot of DHCP renew messages from all the APs. I checked lease time was one whole day, but APs were renewing (also checked on the MS DHCP server logs) too often, and randomly. I suspect APs do not behave well when renewing their IP addresses, and they might drop the terminals associated to them for a few seconds. For 5250 emulation this is pure hell. So I suggested the customer to increase lease time up to infinite. Now everything seems to work much better!! But question remains: why were they renewing so often? Actually... can the customer blame us for this? Are we 100% innocent? Is this a known problem of AP DHCP client? All I could think is: either AP650 were randomly rebooted for some unknown reason (and then request an IP address), either DHCP server was sending unexpected FORCERENEW unicast messages. However, given that after setting lease time to infinite all works better, I do not think of APs being rebooted. I do not think either of MS DHCP server doing weird things... Any ideas?
AP650 DHCP spurious address renewal |
4 Replies
Thanks Kevin! Unfortunately, now all APs have an infinite lease time and they are not renewing their addresses at all. All terminals are working much finer now, so I do not think customer would agree on moving back to a finite lease time just to check what was wrong, and risking getting into trouble again.
Hi Kevin, They are running 5.4.0.0-047R, and yes, traffic is L3 tunneled and actually routed to/from RFS, I should have specified this on my question. Thanks!
This does indeed seem strange. My recomendation would be to view the lease information for the client to determine what the DHCP server is supplying. This will tell you if the issue is WiNG related or DHCP server configuration related. rfs7000-1#service show dhcp-lease on ap6532-1 -------------------------------------------------------------------------------- ITEM VALUE -------------------------------------------------------------------------------- interface vlan21 expire 1 2013/1/14 03:00:24; fixed-address 192.168.21.104; option dhcp-lease-time 86400; option dhcp-message-type 5; option dhcp-server-identifier 192.168.10.6; option domain-name "tmelabs.local"; option domain-name-servers 192.168.10.6; option map-cfg "pool1=192.168.20.24 option map-cfg 192.168.20.25;hello-i..acency-hold-time=15"; option routers 192.168.21.1; option subnet-mask 255.255.255.0; rebind 1 2013/1/14 00:00:24; renew 0 2013/1/13 12:26:47; -------------------------------------------------------------------------------- On my DHCP server I set the default lease time to 24 hours and the max lease time to 48 hours and I don't see this behaviour. Regards, Kevin
Which WiNG release is the customer running? I remember seeing something simular in earlier WiNG 5.2 releases but this has been addressed in later WiNG releases. FYI - The Access Points should not be dropping any traffic unless they are Layer 3 adopted and tunneling the traffic to the Controller. If the traffic is locally bridged you should not be seeing any imopact on the clients. Regards, Kevin