01-16-2017 05:49 AM
Our customer is replacing a pair of Infoblox IB-250 units with a pair of TE-810 units with the same configuration. These connect to a pair of Extreme Blackdiamond switches in two comms rooms, one Infoblox on one switch, the other on the other switch.
Before installing the TE-810s into their racks, we had them on a desk connected to a Netgear switch. We disconnected the IB-250s and connected the Netgear switch to the core Extreme Blackdiamond. In this configuration, everything was working fine, the active unit was issuing IP addresses and resolving DNS queries.
We then shutdown Infoblox 2 and installed it into the comms room and connected it to the Extreme Blackdiamond directly. It was at this point that our troubles started. The unit detected Infoblox 1 as the master so went through the HA restart process, however it never went into the “starting services” state. The only way around this is to power cycle it. We disconnected it from the network and powered it on and the services started. When it was reconnected to the network, it went through the HA restart and the services didn’t start up again.
We put the unit back onto the Netgear switch and it started up fine!
Thinking that there could have been something strange with the Netgear, we put both of them into their final locations on the Extreme switches as described in the first paragraph. Depending on which unit started up first, the second unit would not start its services.
We tried disabling spanning tree on the ports and disabling igmp-snooping on the VLAN where they sit to no avail. A check was made to see that there were no other VRRP conflicts on the VLAN where they sit, the Infoblox’s use group 90, the switches are using group 1.
At no point are the old and new units connected to the network at the same time. The new units are running version 7.3.8. I'm pulling what's left of my hair out over this. Has anybody had any similar experience? I shall be re-visiting the customer next month so any help will be great.
01-17-2017 01:34 AM
HA restart mostly happens because of the config in switch. Try opening a support ticket I am sure they have this reported and fixed.