cc_staff
318
edits
Handling large collections of files: Difference between revisions
Jump to navigation
Jump to search
Handling large collections of files (view source)
Revision as of 17:06, 21 February 2020
, 4 years ago→RAM disk
No edit summary |
|||
| Line 61: | Line 61: | ||
The <code>/tmp</code> file system can be used as a RAM disk on the compute nodes. It is implemented using [https://en.wikipedia.org/wiki/Tmpfs tmpfs]. Here is more information | The <code>/tmp</code> file system can be used as a RAM disk on the compute nodes. It is implemented using [https://en.wikipedia.org/wiki/Tmpfs tmpfs]. Here is more information | ||
* <code>/tmp</code> is <code>tmpfs</code> on all clusters | * <code>/tmp</code> is <code>tmpfs</code> on all clusters | ||
* /tmp is cleared at job end | * <code>/tmp</code> is cleared at job end | ||
* like all of a job's other memory use, falls under the cgroup limit corresponding to the sbatch request | * like all of a job's other memory use, falls under the cgroup limit corresponding to the sbatch request | ||
* we set the tmpfs size via mount options at 100%, which could potentially confuse some scripts, since it means /tmp's size is shown as the node's MemTotal. For example, df reports /tmp size as the physical RAM size, which does not correspond to the sbatch request | * we set the tmpfs size via mount options at 100%, which could potentially confuse some scripts, since it means /tmp's size is shown as the node's MemTotal. For example, df reports /tmp size as the physical RAM size, which does not correspond to the sbatch request | ||