[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] flocking

Kewley, J (John) wrote:

Absolutely, but its not a connection thing, I've only gotten this far by opening up specific ports bidirectionally and specifying hi and low port range of those open ports. This is different than the connection errors I got before... (didn't save any of em, but you get the point).

Some more questions (in the presence of firewalls, the following is important to check, but I don't think is causing your problem):

1. Have you opened 9614, 9618 and your lowport/highport range on the
firewalls for
udp AND tcp on your central nodes and all your execute and submit nodes
in both pools ?

Yes, and I'll doublecheck that with our network guys (our firewall is managed at a different facility). UDP may have been ignored in that, and that concerns me.

2. Are the nodes of either machine on a private network ?

We are all on the same private net (hence 10.*).
I'm reasonably sure that the problems that you're leaning towards here would have different symptoms... this seems to be *almost* a connectivity problem (if a negotiator has to be contacted on the port listed in the flock_from schedd log I posted, and not only reliant on the lists in condor_config).