On Mar 25, 2008, at 12:39 PM, Greg Thain wrote:
So, for schedd failover, the schedd will have a different name, depending on which machine is "live"?
Sounds evil and wrong.
(Assuming we aren't playing games with IP addresses or DHCP). And thus I have to know which schedd is live, and pass that on the command line to condor_q and friends?
Also evil.
(I assume that personal condors predate schedd failover?)
Of course. This is just one of many examples of things that no one bothered to think through when the HA stuff was first going into Condor.
Therefore, yeah, something's going to have to change here to make this feasible in the HA world, since how it all "works" now is obviously not going to fly. I don't have the mental space right now to come up with a full proposal, but I'd be happy to consult with anyone who comes up with such a proposal and shed whatever light I can on it.
Cheers, -Derek