2800AP - VLAN with incorrect router address,- how tollorent?

[craiglebutt]

Hi

So just upgraded 5520 to 8.10.170 from 8.5.182

95 % of the AP upraded no problem, 5% didn't.

I saw them connect to the WLC and upgrade but didn't after that.

What we found was the Router address was incorrect for the vlan in DHCP.  

This mistake must have been there for years when VLAN was created and the APs have been connecting and desktops connecting on the same subnet.

Desktops seem to tollorate the incorrect router and so have the APs untill this latest update, has anything change on the new code to affect this.

It helped fix something that we didn't know was broken.

cheers in advance

[Flavio Miranda]

Hi

 By Router you mean the default-router or default gateway was incorrectly relivered?  There´s no way a device communicate with no gateway unless you have only one vlan and everything is inside this vlan.

[craiglebutt]

HI, it was the default router address in DHCP (Windows) the IP, Subnet and Gateway all correct

[Rich R]

So I presume you were referring to 8.10.171.0 not 8.10.170?

Yes, there were originally undocumented (and VERY stupid) code changes related to gateways on all COS APs that caused a number of problems a while back - it more or less broke our network in places (all APs reloading every 4 hours)!  The change was only ever intended for mesh APs but got applied to all and in a really stupid way initially in 8.10.105.0.  All "fixed" from 8.10.130.0 but still with a permanent behaviour change - the AP must now be able to ARP the gateway IP.

I explained the details in https://community.cisco.com/t5/wireless/ap-1562-map-won-t-join-rap/td-p/4444496

> it was the default router address in DHCP (Windows) the IP, Subnet and Gateway all correct

In DHCP option 3 (Router) *is* the default gateway address so how could gateway be correct if router is wrong as they are one and the same? https://www.iana.org/assignments/bootp-dhcp-parameters/bootp-dhcp-parameters.xhtml

If your gateway was wrong and you used helper address your APs could have managed using broadcast before, or if WLC was on the same subnet, but after the change they poll the "gateway" regardless and that would explain your problem.

I have no idea how your PCs worked without correct gateway without seeing packet captures but maybe they discovered it through ICMP redirects and the APs might even have done the same.

[MHM Cisco World]

can you more elaborate ?