File Transfer with Globus¶
Avoid Direct Transfers via Login Nodes
Please refrain from using scp
, sftp
, rsync
, or direct connections to the login nodes for data transfer. These methods can degrade system performance and are less efficient compared to Globus Connect.
Overview¶
REPACSS supports high-performance data movement through Globus Connect, a robust tool designed to facilitate large-scale file transfers. It provides:
- High-speed, reliable transfers
- Automatic error detection and retry
- Multiple parallel streams for faster throughput
- No impact on login node performance
- User-friendly web-based interface
Setting Up Globus Connect¶
Follow these steps to enable file transfers using Globus Connect Personal:
-
Install Globus Connect Personal on your local machine:
- macOS Installation Guide
-
Create a Personal Collection:
-
After installation, set up a Globus collection tied to your system.
-
Access REPACSS Data:
-
Set the endpoint to:
REPACSS
-
Navigate to the appropriate storage paths:
- Home:
/mnt/GROUPID/home/USERID
- Scratch:
/mnt/GROUPID/scratch/USERID
- Work:
/mnt/GROUPID/work/USERID
- Home:
Transferring Data Between Sites¶
Many academic and research institutions have established Globus endpoints. To transfer data between REPACSS and other sites or to your own machine:
- Use the Globus Web Interface to configure transfers.
- Specify the collection to be used:
- REPACSS Endpoint:
REPACSS
- Locate and select the file(s) you intend to transfer, then click the
Transfer or Sync to...
button. - Select your destination collection.
- Specify the target file path and click Start to initiate the transfer.
Best Practices¶
- Always verify file integrity post-transfer.
- Monitor job completion using the Globus web interface.