You know, VM HA is the thing that I get the most frustrated with with VMware. Which is kinda ironic - is an area where EMC and VMware have a long trail of cooperation (check out the name of the service during startup, or the firewall rule for the VM HA agent if you want to decode what I mean). DNS is super important, make sure you always have that right.
To be fair: it's WAY easier and IMHO much more flexible than the other ways HA is done in VMware's competitors - compared with WSFC (aka MSCS) it's a walk in the park, and offers much more granular object failover (but less granular failure detection). I'm just saying a better job could be done documenting some of these things (something I know others have championed).
I had configured the das.isolation change once before, but then needed to reconfigure it when I was rebuilding my clusters at home, and just couldn't get it to work...
Here's how you do it...
Ok - after adding "das.isolationaddress1" in the VM HA "advanced options" window, I kept getting this screen:
Argh - I know you can't reach the gateway!!! My poor little Linksys router cant understand anything except a class-c /24 subnet - that's why I specified das.isolationaddress#!
WHY DO I KEEP GETTING THE ERROR!
Ah, phew - finally found the original KB article here (definitely bookmarkable). You have to also specify: das.usedefaultisolationaddress=false
So, here's where you make the changes (note how LAN and iSCSI are on seperate subnets, and I've got two das.isolation addresses - one on each - you should also have other vmkernel traffic like vmotion of a different subnet also):
Mental note to self..... Someone should really post all the non-documented advanced options.... Has anyone seen this anywhere? Otherwise, I'll ping my VMware compatriots.
Ok - now that the cluster is happy again, I can focus on something fun, not something so pedantic. I'm going to do a series on "HOWTO _____ using the Celerra Simulator"....