Controllerless Networks

I am having trouble with an IAP-225/IAP277 mesh configuration.

Firmware Version - 

I am able to bring all of the IAPs up in the cluster and apply the default settings.

I create my SSIDs and turn off extended SSID.

 I then select the point IAP and set its uplink to Ethernet Bridging enabled. 

Then I move the IAP to a remote MAS switch and power it up over PoE. 

A mesh link is created and I can ping the portal and the point for about 1 minute but then the two drop from the original cluster and it's almost as if they are trying to create their own cluster. I completely lose connection over our network to both the point and the portal at this point.

When I console into the two IAPs they display eachother as point and portal, but again, they have lost all connection to the original cluster/rest of the LAN.

Any suggestions?

The mas on the point side is completely isolated and not connected to any other network, and there's nothing handing out DHCP on the mas/point side? Are you initially standing this up all on one flat VLAN I assume?

That is correct. The MAS switch is isolated, as the Bridge is intended to be the connection to our existing network. 

I have tried this on a completely flat lab network as well as our production VLAN'd network. I am seeing the same results on both environments.

When you bring them up, capture a show tech from both while they are up and then again while they are down. We are sure there's no duplicate IPs, no port security on the portal's uplink into the switch etc?

You can private message them to me (both before and after). Also make sure that the port that the portal is connected to, on the switch, sees two mac addresses (one should be the portal's mac, the other the point's mac). 

We had a similar issue with AP-277 in a mesh.

Make sure the PoE power supplied to the AP-277 is sufficent.  We had similar issues with the APs booting up ok, but not enough power to get the radios going.  A PoE+ injector would be best.  Note that if you are using a injector, ensure the switch port PoE power is disabled as well otherwise the injectors own power will not be supplied to the AP.

If you don't have an injector and the switch is a HP\Aruba switch. Update the switch firmware to the latest and if required, enter the following commands on the switch port:

power-over-ethernet high
poe-allocate-by value
poe-lldp-detect enabled

poe-value 25

Thank you for the suggestion, FabianFlorian. Unfortunately, the power settings did not fix the issue. I have also tried the setup in my lab just using two IAP-225's (which can definetely be fully powered by the MAS switches) and I am seeing the same symptoms.

Once the mesh point and portal find eachother they stay connected via the radios, but neither can contact the rest of the network.

I'll give it another go.

1 - Make sure the AP that is the Portal function is the preferred Master

2 - Only the Point AP should be Eth0 Bridged

3 - Make sure the Switch port vlan assignments tagged\untagged are aligned with the AP Eth0 and Eth1 port Wired Network assignments on both APs.  Note the Wired Assignment settings are across all APs and there ethx ports in the cluster. So if you are running a trunk  (native + tagged vlan(s)) on the portal end Eth0, the same switch port configuration is required to connect to the point Eth0 port.  Also make sure you "allow" all the vlans on a trunk mode port ;)

4 - Set static IP addresses for the APs and VC

PROGRESS!

Setting the Portal to the preferred master actually got things going! I didn't see it noted anywhere that the Portal needed to be the master, though! Does this mean that my bridge will go down if the master switches to another IAP?

Thank you for your help on this!

I'm not qualified to answer that question.  Best to talk to your local HP Business Development Engineer .  But logically, if your Portal AP fails, the mesh will be down or isolated from the main network.