[Condor-users] CCB, glidein, and NAT

[Peter Doherty <doherty@xxxxxxxxxxxxxxxxxxx>]

Hi all,

I'm using a setup involving Condor glideins and CCB.  The collector  
has a 1:1 NAT configuration with an private IP, and a public IP that  
has all traffic forwarded to it at the router.

I have a collector/schedd submitting jobs to external pools, and the  
glideins (the startd) are falling to call home to the collector.
Somewhere the startds have been giving the collector's private IP,  
instead of just trying to contact it by hostname, which would work find.

Is this something that is within the design of Condor CCB, and I just  
need to set the correct variable in the CONDOR_CONFIG to make this work?
I've set on the collector and schedd:
CCB_ADDRESS = collector.hostname.com (as an example)
I can access the external pools directory, and I checked their config  
files, and CCB_ADDRESS is set correctly.


here's the log from one of the Startds that couldn't call home:
05/06 17:25:46 (pid:15513) CCBListener: registered with CCB server  
collector.hostname.com as ccbid 10.0.10.41:9618#1493
05/06 17:35:27 (pid:15514) attempt to connect to <10.0.10.41:48744>  
failed: Connection timed out (connect errno = 110).  Will keep trying  
for 300 total seconds (111 to go).
05/06 17:37:19 (pid:15514) attempt to connect to <10.0.10.41:48744>  
failed: Connection timed out (connect errno = 110).
05/06 17:37:19 (pid:15514) CCBListener: failed to create reversed  
connection for request id 208 to <10.0.10.41:48744>: failed to connect


One of the pools I'm submitting to is on the same network segment, and  
everything works fine with that pool.  So I feel pretty confident in  
saying that the only problem is just that the startds have all been  
given an IP that they shouldn't be using, instead of being given a  
hostname that they can resolve and connect to.

Thanks for any advice.

Peter