Mailing List Archives
Authenticated access
|
|
|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Condor-users] Condor flocking
- Date: Mon, 6 Feb 2006 19:01:05 -0000
- From: "Kewley, J \(John\)" <j.kewley@xxxxxxxx>
- Subject: Re: [Condor-users] Condor flocking
> Dear Condor users,
>
> I'd like to know if Condor flocking is only allowed if all machines in
> different pools (that are going to flock jobs) have visible IP
> addresses.
>
> I looked at various sources (condor manual, faqs and google search)
> and it seemed to me that all machines involved in flocking need to
> have a visible IP number to be placed as a value for the variable
> FLOCK_FROM in the condor_config file.
Worse than that:
Any potential execute node must be able to communicate with any possible
submit node for the allocated high port ranges over both UDP and TCP,
and I believe vice-versa. Firewalls at submit and execute nodes and any
firewalls in between must be open to allow this traffic.
The situation does not change when flocking is involved - matched nodes need to
talk directly, comms is not directed through central managers.
See
http://www.nesc.ac.uk/talks/ahm2005/431.ppt
Paper should also be at
http://www.allhands.org.uk/2005/proceedings/papers/431.pdf
but doesn't seem to be there just now, if it doesn't come back soon,
email me and I'll send you a copy of the paper.
Cheers
JK