a week ago
There is an issue I've come across that I've opened a case for, but wanted to start some technical discussion on. Bare with me as I want to be detailed in describing my situation.
- At times the IAP virtual controller IP address will become unresponsive for a cluster.
- Trying to access the GUI will fail, and pings to the VC IP address will fail.
- Pings to the subnet default gateway are successful
- No IP conflict issue & VC IP address is removed from DHCP scope
- Pings from gateway to VC IP address fail (of course, no ARP entry at switch with default gateway)
- Pings from an AP within the cluster succeed
- result in ARP entry at switch with default gateway
- pings from default gateway to VC IP address are now successful
- pings from external network still fail
- Clear arp table on switch and go back to pings from gateway to VC IP address fail
- Reboot of master AP did not resolve the issue
- Reboot of AP cluster resolved the issue
This ultimately seems like a ARP and/or routing issue (reaching external networks via the default gateway). The current design does not mirror Aruba recommendations (AP management VLAN) and we are making changes to do so. However, how can this guarantee that these same issues don't appear when we make these changes? I've verified all configuration, and it is correct (Uplink vlan, switch (ProCurve) port configuration, VC IP configuration).
Sorry, I don't have packet captures as this was a remote site from where I am at.
a week ago
Could you give us some more information about your cluster? What type of IAPs, code version running, size of cluster etc?
I've seen issues with VC response in clusters that are very large (ie. over ~80 members) and have a lot of multicast traffic riding the network.