[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Condor-users] Globus jobs failing using Condor-G
- Date: 29 Sep 2004 11:24:28 -0500
- From: Carey Kireyev <ckireyev@xxxxxxxxxxx>
- Subject: Re: [Condor-users] Globus jobs failing using Condor-G
i suspect the problem is that the gatekeeper node cannot talk back to
the client (Condor-G) machine (due to the firewall).
To get more specific information, do the following:
1) set the GRIDMANAGER_DEBUG = D_FULLDEBUG in your condor-g
2) re-submit the job
3) in your gridmanger log file (GRIDMANAGER_LOG, usually =
/tmp/GridmanagerLog.<user>), look at the RSL clause, e.g.:
9/28 17:57:15  Final RSL:
The GRIDMANAGER_GASS_URL (e.g. https://vdt-redhat9.cs.wisc.edu:57943) is
the address on the client (Condor-G machine) that the gatekeeper tries
to connect to but probably cannot in your case.
To test, you can ssh into the gatekeeper machine and use telnet to
connect to that address. (If everything's ok you should get "connection
closed by host". If the server can't connect, you should get "connection
refused" / "connection timeout", etc).
If that's not the problem check to make sure that the clock skew between
server and client is not too great.
On Wed, 2004-09-29 at 10:09, A Nayar wrote:
> I have successfully tested submitting globus jobs to the teragrid through
> condor-g in one of our pools but in the other pool which is firewalled I
> could not, I opened up all the ports that globus is supposed to use but still
> get the error 74: failed to open stderr
> I can do a grid-proxy-init allright but no globus-job-run. Also I just
> installed the resource management bundle because on condor-g install page it
> said thats all that is needed. Any help would be appreciated