There's a pretty comprehensive treatment of the subject at a PDF at
http://www.globus.org/toolkit/security/firewalls/. Start there, and
see if there are still questions left.
Charles
On Mar 12, 2008, at 5:23 PM, Nyarfee wrote:
Hi:
Well, I've set the GLOBUS_TCP_PORT_RANGE with the open range in the
file /etc/xinetd.d/gsigatekeeper on the client side (my machine)
and in the server (our cluster), but there is a firewall that
manages (open/close) the ports in the "domain" were my machine and
our cluster are, so I just have to open the port range and the port
of the gatekeeper (of my machine and our cluster) in the firewall?
or I have to open it in my machine and in our cluster too?
Thanks and Regards...
2008/3/10, Charles Bacon <[EMAIL PROTECTED]>: That means the
gatekeeper cannot contact your client. If you're
behind a firewall, you'll need to open a port range to be contacted
on, and set GLOBUS_TCP_PORT_RANGE on the client side to the open
range. If your machine has a private address, but can be reached via
NAT or the like, you need to set GLOBUS_HOSTNAME on the client side
to the correct hostname for the server to use to contact the client.
Charles
On Mar 10, 2008, at 2:34 PM, Nyarfee wrote:
> Hi:
>
> I'm verifying Globus Installation with the steps from http://
> www.gridway.org/documentation/stable/installconfguide/x108.htm, the
> commands works well in the machine where I installed GT4.0.5
> (localhost) except step 3. Submission test in the
> name_of_our_cluster:
>
> 1. Authorization test:
>
> $ globusrun -a -r localhost
> $ globusrun -a -r name_of_my_machine
> $ globusrun -a -r name_of_our_cluster
>
> I receive in all: GRAM Authentication test successful
>
> 2. File transfer test:
>
> $ globus-url-copy file:///etc/hosts gsiftp://localhost/tmp/hosts1
> $ globus-url-copy gsiftp://localhost/tmp/hosts1 file:///tmp/
hosts2
>
> $ globus-url-copy file:///etc/hosts gsiftp://name_of_my_machine/
> tmp/hosts1
> $ globus-url-copy gsiftp://name_of_my_machine/tmp/hosts1 file:///
> tmp/hosts2
>
> $ globus-url-copy file:///etc/hosts gsiftp://name_of_our_cluster/
> tmp/hosts1
> $ globus-url-copy gsiftp://name_of_my_machine/tmp/hosts1 file:///
> tmp/hosts2
>
> All transfer are ok.
>
> 3. Submission test:
> $ globus-job-run localhost /bin/uname -a
> $ globus-job-run name_of_my_machine /bin/uname -a
>
> the outputs are ok. but when I type
>
----------------------------------------------------------------------
> ---------------------------------
> $ globus-job-run name_of_our_cluster /bin/uname -a
>
> this message appear: GRAM Job Submisison failed because the job
> manager failed to open stderr (error code 74)
>
----------------------------------------------------------------------
> ----------------------------------
> The cluster have GT4 installed and configured Pre-ws, I already
> have the Certificates Authorities and I create a valid proxy before
> type the commands. The GT4.0.5 in my machine is just installed and
> configure to work with the meta-scheduler GridWay 4.2.3 (and I
> don't need GT4.0.5 in my machine like one node worker).
>
> What could be the reason of this message and how I can fixed?
>
> Regards...
