Storage¶

Hydra has four kinds of disk space. Keep scripts and configuration in /home, results in /data, the working files of jobs in /scratch, and project data waiting for analysis in /store. Each has its own quota and retention rules. See Filesystems.

To see how much space you are using, check your usage and quotas.
To get back a file you deleted or overwrote on /home, /data or /store, recover it from a snapshot.
To get back a file the scrubber removed from /scratch, request a restore within about ten days.
For a job that reads and writes intensively, use the node's local SSD.
For data on /store, copy it with the I/O queue before a job uses it.

Files on /scratch/public older than 180 days are deleted every week

/scratch has no snapshots and no backup. Keep results on /data or off Hydra. We can restore a scrubbed file for about ten days after the scrubber removes it.

None of the storage systems on Hydra are meant for archival storage. Backups explains what is protected against what, so that you can decide where else to keep a copy. To move data on and off the cluster, see Data transfer.

Last updated 2026-10-03