Using Globus to Transfer Files¶
If you are looking for instructions on using the globus-cli command line tool, please visit the globus-cli Command Line Tool page.
Why use Globus?¶
- Globus is the fastest and most effecient file transfer option, especially for large transfers to or from PACE clusters
- PACE provides free access to the GT community
- Has a graphical web-based interface for ease of use
- Provides the option to encrypt files
- Automatically resumes file transfers that are interrupted
Step 1: Logging In¶
- You can authenticate to Globus with your Georgia Tech credentials, so there's no need for a separate account
- Go to https://www.globus.org/ and click Log In
- Search for and select "Georgia Institute of Technology", then use the GT login page
- After logging in with your GT credentials, you will be redirected to the main file manager screen
Step 2: Installing Globus Connect Personal¶
- If you'd like to transfer files to or from a personal or local computer, you'll need to install Globus Connect Personal on the computer. If you're moving files between PACE clusters or between PACE and other research computing sites that use Globus (such as an XSEDE resource or national lab cluster), then the local client is not needed, and you can skip to step 3.
- The Globus Connect Personal client allows Globus to access files on your local computer or transfer files to it.
- After installing the personal client, you will set up your computer as a collection to allow files to be transfered to and from it
Log in to Globus then follow the appropriate guide on how to install Globus Connect Personal and set up a personal endpoint:
The above guides should also cover how to setup your personal endpoint (also referred to as a collection). A personal endpoint is simply your personal computer that you want to share files from. You should see a screen such as the one below during the globus connect personal setup:
- Tip: By default, Globus will only have access to certain folders on your computer (usually only
/documentson Windows or your home directory on Mac). To make more folders accesible to Globus on Windows, right click the Globus Connect Personal icon, then select options and click the
+sign to add folder paths. On Mac, click the icon, then Preferences, then select the Access menu to adjust what Globus can see.
- Linux is a bit tricky, but it is done most easily using the graphical interface. After using tar to extract the files, if
./globusconnect & gives errors about tcllib, skip to the part of the guide titled "How to Install Globus Connect Personal for Linux Using the Command Line" and follow that to install without the graphical interface. Try
./globusconnectpersonal -start & if
./globusconnect & doesnt work
- Linux: Globus also has a standalone command line tool, globus-cli, which is separate from Globus Connect Personal.
Step 3: Set Endpoints for Transfer¶
Make sure the Globus Personal Client is running on your local computer, if you will be transferring files to or from it.
- Log into Globus, and you will be taken to the main file manager screen
- To set up your local computer running Globus Connect Personal as one endpoint:
- Click on the Collection Bar at the top
- Click on "Your Collections"
- if you set up your endpoint correctly following the documentation linked above, the endpoint you created should show up
- select it and you should be taken back to the file manager page, with your computer's files available to transfer showing up on the left
- Next, set the cluster as the other endpoint:
- Click "Transfer or Sync to" which should shift the panel view to the right
- Select "collection" bar on the right (next to the collection bar you just used on the left)
- Search for "PACE"
- PACE Phoenix: For file transfer to/from the Phoenix cluster, select the 'PACE Phoenix' tile which should be green
- PACE Hive: For file transfer to/from the Hive cluster, select the 'PACE Hive' tile which should also be green
- PACE Phoenix VAPOR NETWORK ONLY: (requires access permission) For file transfer between the Phoenix cluster and campus research computers connected to the Georgia Tech internal high-speed Vapor network, select this option. Note that transfers will fail if the data is not originating from a Vapor-connected computer. Contact PACE Support for more information and to request access to this endpoint.
- You will be prompted for a authentication, with the domains where you have access. Click on Continue
- After you click continue, options for identification will be displayed. Select your GT identity (or GT firstname.lastname@example.org) to continue with your PACE credentials.
You will be presented with the usual GT authentication. Follow the prompts and enter your password and duo information. Select Allow for the prompts.
Your cluster account and all your files on the cluster should now show up, ready for file transfer. You will be in your home directory, and you can navigate from there in the folder tree or use the Path bar above to reach the directory you wish to use.
- If you wish to transfer between two clusters, there is no need for a local endpoint. Just select each cluster on one side of the file manager and tranfer in the same way. You can transfer between PACE Phoenix and PACE Hive or between PACE and another computing site, such as an XSEDE resource. The collection search bar can be used to find the other cluster.
After the first time setting the endpoints, they will be saved under recent, so you can easily select them. The Globus Connect Personal client must be running before setting your local computer as an endpoint.
Step 4: Transfer files and folders¶
You can now transfer files between your account on the cluster and your personal computer, in either direction. Simply check the files and/or directories you wish to transfer from the source, ensure that the destination is set to the directory where you want the files to arrive, and select the Start button in the proper direction.
Select the folder / file you want to transfer
Transfer or Sync to...on the menu in the middle
- Select the folder you want to transfer to
- Select one of the
Startbutton at the bottom of the screen, with the arrow direction corresponding to the direction of transfer, to initiate the transfer
- Detailed status of the transfer can be viewed in
Activitylocated in the menu on the left (click on top left corner if it is not shown)
- Globus should send you an email when the transfer is complete. It will automatically restart and continue if there are any disruptions along the way, making it easy to transfer large files that may take time.
- Example: Here I have selected the
win-libraryfolder on my laptop and my
datafolder on the cluster
- Transfer from my laptop to the cluster: To transfer the
win-libraryfolder to the cluster, I would select the start arrow in the bottom left
- Transfer from the cluster to my laptop: To transfer my
datafolder from the cluster to my laptop, I would select the start arrow in the bottom right
Using Globus Recap¶
Always make sure the Globus Personal Client is running before you go to transfer files to or from your local computer.
- Start Globus Connect Personal if transferring to or from a local computer.
- On Linux, navigate to where you installed the globuspersonalclient files,
cd globuspersonalclient-x.y.z (x.y.z is version number)and then run
./globusconnectpersonal -start &
- On Linux, navigate to where you installed the globuspersonalclient files,
- Log into Globus, which directs you to the main file manager page. Other pages including activity page and endpoint manager are available in the drop down menu on the left (click top left if the menu is not visible)
- To set your personal machine as an endpoint, click on the "collection" bar on the top left (on file manager page), select your personal computer name (personal endpoint name you made). It should be under the "Your Collections" menu. To set the cluster as the other endpoint, click on the collection bar on the top right, search for PACE, and select
PACE Hive. Log in with your GT username and password. Endpoints should also show up in your recent endpoints.
- You can now transfer files and folders as you wish. Monitor transfer status with the
- When you are done transfering files, logout of globus online, and logout of the personal client on you computer. For linux, logout by running
Sharing Files with Globus¶
Globus can be employed to easily share files with collaborators, whether within or beyond Georgia Tech, without requiring a PACE account for the collaborator. Find out more at How To Share Data Using Globus. The collaborator will need to be able to log into Globus; with many universities, researchers can do this with their own university login (as GT offers with Globus), but there are options for Google accounts or for getting a Globus-specific account if needed as well. Files or directories can be shared from Phoenix or Hive. Sharing from a personal endpoint is also possible after you request to join the Globus Plus group, as described below.
Globus Plus Features¶
Georgia Tech's Globus subscription, managed by PACE, allows members of the GT community to request access to enhanced features available with Globus Plus by joining the Georgia Tech Campus group. Most users do not need to join this group. Transferring data to/from PACE resources and to/from/between other computing centers with Globus subscriptions does not require the enhanced features.
Georgia Tech researchers who wish to use Globus for transfers between two personal endpoints or to use sharing on personal endpoints (not on PACE) should request to join the "Georgia Tech Campus" group. From the Globus File Manager, click Groups on the left menu, then search for the group by name and click the "Join Group" button. You should also send a short explanation of why you need access to the group to PACE Support. A PACE administrator will review your request and contact you for additional information.