Good afternoon, I have been directed to install a Grid FTP server to allow large data transfers to happen much quicker than standard SCP. I am installing on CentOS 6.2. After a several day nightmare, troubleshooting vague error messages where every *single* step took hours to complete, I think I finally have things installed correctly. For the most part I followed the directions here: http://globus.org/toolkit/docs/5.2/5.2.2/admin/install/ and here: http://www.globus.org/toolkit/data/gridftp/quickstart.html I am using the CentO packages, with self-signed certificates.
On the server I am issuing the following commands: sudo service globus-gridftp-server start sudo grid-proxy-init -verify -debug -valid 4:00 grid-proxy-init -verify -debug -valid 4:00 Then on the client I issue the following command: time globus-url-copy -v file:/opt/120_GB_file.random sshftp://172.22.10.206/home/test/bigfile.random Which returns: real 17m52.865s user 1m55.064s sys 1m50.117s As you can see, the transfer just under 18 minutes. Then I issue the following command: time scp /opt/120_GB_file.random [email protected]:/home/test/bigfile.random Which returns: 120_GB_file.random 100% 111GB 106.9MB/s 17:48 real 17m52.355s user 7m31.879s sys 5m40.904s As you can see, the second transfer took half a second less using scp vs. gridftp. Is this the expected result? If not, can anyone start me along a path of investigation that might prove useful? For instance, how can I log what is happening? In addition, is there a single, from start to finish, simplest case tutorial? The documentation that is available starts with the premise that the reader is setting up their production environment right from the start. I am just trying to set up the absolute simplest possible test environment between two servers that are on top of each other in the same rack, so I can get an understanding of what is happening before opening service and ports to the outside world. Thanks, Melvin
