What is CAC Archival Storage?
- CAC Archival Storage provides low-cost, high-performance option for storing research data.
- CAC Archival Storage has a direct 100 Gbps connection to Internet2.
- CAC Archival Storage is not mountable by running jobs, instead the user must transfer their data from the CAC Archival Storage to an accessible server using Globus.
- Globus users have easy access to add, delete, and share their data using any Globus endpoints.
- Some of the Globus endpoints available include:
- storage01.cac.cornell.edu (cac#home) where all CAC user home directories are found
- XSEDE sites:
- Stampede (xsede#stampede)
- Lonestar (xsede#lonestar4)
- TACC Archival Storage (xsede#ranch).
First step - Enable (or create) CAC project for Archival Storage and add users where appropriate
Second step - create your Globus account
When signing up with Globus, Cornell users should select Cornell University under Use your existing organizational login and then click on the Continue button. You will get forwarded to the CUWebLogin page. Login using your Cornell NetID and password.
Globus can be accessed using the following methods:
- Globus Web GUI: click on the "Login" button on the Globus web site to start.
- Globus CLI client on your computer:
- Globus SDK for Python: for workflow automation and integration with your science gateways or third party software.
Note: The legacy ssh-based hosted CLI will be deprecated in the future. Please do not use it for new development. If you use the ssh-based CLI for current production, you will need to migrate to use new Globus CLI client or Globus SDK for Python soon.
CAC's EndPoint is CAC Archive 2/DTN (cac#archive02).
- When activating cac#archive02 endpoint in Globus web GUI, you will be prompted by a dialog box saying:
The administrator of this endpoint, cac#archive02, requires that you authenticate using their MyProxy OAuth server to activate the endpoint. When you click 'Continue' you will be redirected to their website.
- You will be redirected to the https://archive01.cac.cornell.edu/oath/authorize... page.
- Enter your CAC credentials.
- When login is successful, you will be redirected back to Globus web GUI with the endpoint activated.
- cac#archive02's default path is /export.
- Each project with access to CAC Archival Storage has a shared directory (named the project) in which all project members have full read/write access.
- Users can rename and move files and directories within their project directory on the endpoint. Globus added this feature recently.
Advanced Topic - Automating transfers to the Archival Storage
- Install Globus Connect Personal on the Linux/MacOS/Windows host you wish to archive by clicking on the "Get Globus Connect Personal" link on the Transfer Files screen on Globus.
- On the host you wish to archive, download and untar Media:archive_scripts.tar.gz.
- To enable running Globus Connect Personal as root, add
- to globusconnectpersonal-2.0.3/gc.py (on line ~ 360):
args = [os.path.basename(PDEATH_LAUNCH), GRIDFTP_SERVER, "-allow-root", "-i", "-always-send-markers", "-hostname", "127.0.0.1",
- Copy root-bin directory from the archive_scripts.tar.gz to /root/bin. If you are archiving directories outside /home, modify the -restrict-path argument in /root/bin/gc_start.sh.
- Generate a ssh key pair using the "ssh-keygen" command, leave private key in ~/.ssh, and upload the private key to Globus
- Make sure you can access Globus CLI like this:
ssh -i .ssh/<private key> <globus user name>@cli.globusonline.org
- Modify archive.sh to match your Globus user name, private key file name, CAC project and archive directory.
- On Globus, make sure your connection to cac#archive02 endpoint is activated.
- You should now be able to run archive.sh to upload your archive directory to CAC archive. You can automate this script using cron.
Advanced Topic: Syncing to Archival Storage
See here for how to sync data to Archival Storage