Controllerless Networks

Reply
New Contributor

New default IAP overrules configured cluster

Hi

We have an established install of over 600 locations each with it's own IAP cluster using MAS or 2530 switches along with lldp for auto port configuration, dhcp options 60 and 43 for amp management.

 

the switch ports are configured with an access vlan for the wired LAN out at each location, upon connection the iap get's its management vlan thanks to lldp.

 

This setup has worked great until we were required to activate dhcp on the access vlan. Even though we've tuned the lease time down to the minimum possible of 5 minutes it has given us major headaches when an unconfigured iap is introduced.

this is what happens

1. iap gets dhcp ip from the access vlan

2. switch reconfigures port thanks to lldp to the correct trunk with management vlan for the iap, however the iap will, until it renews it's IP have an incorrect ip in the correct vlan.

3. after 2,5 minutes the iap gets it's correct IP in the management vlan, by this time the new iap has booted, it can overule the already configured instant cluster, come in in to amp as a new device, start a reboot cycle or various other symptoms, the result is anarchy.

4. the correct instant profile includes a newer firmware so we can also see rebootes connected with that and as the correct config profile hasn't been installed it can also add to the instant chaos.

5. firmware version used is 6.4.2.6-4.1.1.10_51810

6. we've tested enabling 'preferred master' this stops the configured instant master from being overruled but the 'new' iap's just ignore the cluster and do their onw thing coming in as a new device instant

7. The only workaround we've managed so far is to turn off poe on the affected iap switch ports, change the port configuration so that the management vlan is the access vlan, then the new iap's come in as normal and join their correct cluster.

 

The question is. Is there a way to get new iap's that have transisted through another vlan/dhcp ip, to join their correct cluster once they can comminucate with the right master?

Has anyone else experience of the same type of scenario?

 

It seems like a factory default instant has a higher priority than an already configured instant. Had lldp been faster than dhcp it wouldn't be a problem. We've even seen situations where, after attempts to fix the problem via amp iap's simply stop trying to get an address from dhcp and the only option is a factory reset.

 

I hope i've managed to describe the problem well enough

 

Thanks in advance for any advice

 

Simon

Search Airheads
cancel
Showing results for 
Search instead for 
Did you mean: