Large data paths

Where a large dataset lives on a hub, so that a tool session can read it without every member carrying a copy in their home directory.

Asking for space

Large data space is granted on request, not by default. Agreeing to it is a decision for the people who run the hub: how much disk is left, how big the dataset is, and whether it belongs on the hub at all. Settle that first, then open a support ticket asking for the directory to be created.

A directory for a tool

/data/tools/<toolname>

Mode 775, writable by the tool's development group, readable by everyone else. Data kept here is meant to be read by that tool while it runs.

The development group is app-<toolname> — com_tools creates it when the tool registration is saved, naming it from the Dev group prefix option (default app-) and the tool alias in lower case, and sets its membership from the development team named on the form. That much is CMS-side and verified. Whether the group grants write access on the shared filesystem is the platform's business.

A directory for a group

/data/groups/<groupname>

Mode 775, writable by the members of that hub group. Data kept here is maintained by the group and can be read by more than one tool.

Reaching it from a session

Both /data/tools and /data/groups are mounted in tool containers, so a tool reads them as ordinary paths. Nothing is copied into the session and nothing counts against the member's home directory quota.

For the other directories a tool session sees, see Tool paths.

Launching a tool at a large data path

A tool session can be started with file and directory names in the launch URL's params argument. com_tools refuses any path that does not begin with a directory on the Directory Parameter Whitelist, and the shipped default lists only /home. So pointing a launch link at something under /data requires an administrator to add that prefix to the whitelist first. See Directory parameter whitelist for the check and how to change it.