Storage
(Last update: Apr 17, 2024)
Disk Spaces
Following disk spaces are available. Please use appropriate storage area for your purpose. For data that you want to keep for a long time, please save it in /home.
| name | backup | retention period | quota | usage |
| /home | no | at least 1 year after the end of use | YES | Data repository on lustre. Currently, there are no functional differences between /home and /save. |
| /save | ||||
| /gwork | no | basically only during running job | no | Temporary space for jobs on lustre. |
| /lwork | no | only during running job | YES | Scratch space for jobs on computation nodes. /lwork/users/${USER}/${PBS_JOBID} is created. This directory is immediately removed after the job termination. |
(The retention period of /home and /save can be extended if there is large available storage space.)
What will happen if exceeded
If the usage of /home and /save exceeds the limit, you and your group member can't submit new jobs anymore. Current disk usage can be checked with showlim command. You can also request extra disk space. If the usage becomes within the limit (and wait for minutes), you will be able to submit jobs again.
If the usage of /lwork exceeds the limit, the job will probably crash. You can increase the available /lwork space by increasing number of cpu cores, since the available space of /lwork is proportional to number of CPU cores.
For insufficient space of /lwork problem in g16sub and g09sub, you can add -N option to avoid the use of /lwork. (/gwork will be used instead. I/O performance of /gwork is not as good as /lwork, though.)