Since the remote grid services on nodeB will need to authenticate to the globus-gridftp-server on nodeA using the credentials for user jane, we need on nodeA to create a grid-mapfile and put user jane's credentials in it. You can do this in the same way that you added the credentials to the grid-mapfile on nodeB.

On nodeA, as user jane run the following command:

[jane@nodeA ~]$ grid-cert-info -subject
/O=Grid/OU=GlobusTest/OU=ps.univa.com/CN=Jane User

As user globus add this user to the grid-mapfile so that the cat command will produce a result like the one below. Be sure to put the certificate name in double quotes as shown, and be sure to add the UNIX account name after the certificate name as shown:

[globus@nodeA ~]$ cat /opt/globus-4.0.1/etc/grid-mapfile

"/O=Grid/OU=GlobusTest/OU=ps.univa.com/CN=Jane User" jane

Now as user root you need to start the globus-gridftp-server on nodeA. Start by defining the GRIDMAP environment variable:

[root@nodeA ~]# export GRIDMAP=/opt/globus-4.0.1/etc/grid-mapfile

Now start the server on nodeA:

[root@nodeA ~]# /opt/globus-4.0.1/sbin/globus-gridftp-server -p 2811 -S

As user jane test the server by seeing if you can move a file from nodeA to nodeB using a 3rd party transfer (a transfer from one server to another, ie. using two gsiftp:// URLs)

[jane@nodeA ~]$ globus-url-copy -vb gsiftp://nodea.ps.univa.com/etc/issue gsiftp://nodeb.ps.univa.com/tmp/foo
Source: gsiftp://nodea.ps.univa.com/etc/
Dest: gsiftp://nodeb.ps.univa.com/tmp/
issue -> foo

Did the transfer work? We can find out without leaving nodeA by running a job on nodeB using GRAM WS:

[jane@nodeA ~]$ globusrun-ws -submit -streaming -F https://nodeb.ps.univa.com:8443/wsrf/services/ManagedJobFactoryService -c /bin/cat /tmp/foo
Delegating user credentials...Done.
Submitting job...Done.
Job ID: uuid:9dc7a3a8-a4a5-11da-9643-0011d8b1eb22
Termination time: 02/24/2006 19:50 GMT
Current job state: Active
Current job state: CleanUp-Hold
Fedora Core release 4 (Stentz)
Kernel \r on an \m

Current job state: CleanUp
Current job state: Done
Destroying job...Done.
Cleaning up any delegated credentials...Done.

So yes, the file was transferred. We are now ready to try running a job on nodeB from nodeA involving the transfer of files from one machine to the other.

< Back Next >