Properties of Lustre files and directories can be controlled or queried through a command-line interface. Among these Lustre File System (lfs) commands, perhaps the most useful is lfs quota, which simply tells you your storage limit on a given Lustre system. Relevant file systems to query at TACC are $HOME and $WORK:

Typical user limits for these partitions on TACC systems are 10-25 GB and 1 TB, respectively.

If you work with extra-large files, say over 100 GB in size, then lfs setstripe is another good command to know. It takes four arguments:

1. File or directory for which to set the stripe.
2. Size in bytes on each OST, with k for KB, m or M for MB, g or G for GB.
3. OST index of the first stripe (-1 for filesystem default).
4. Count or number of OSTs to stripe over.

As a general guideline, you should consider setting a stripe for every 100 GB in the file. For example, to stripe an intended 200 GB file across two OSTs, create the file by calling:

When you omit the -S argument in the above, you are setting the stripe size to be the system default, which is 1 MB at TACC. You needn't set it any lower than that (and you can't set it lower than 64k). But if your application uses parallel I/O with large files, you might obtain better performance if you set a larger stripe size on your new output file(s). For example:

To see Lustre's default striping properties, try creating a small file and then using lfs getstripe to get its stripe information.

The listing at the end of the results shows which OSTs have parts of the file.