Sync behaviour¶

RAGflow runs the connector on a schedule for the knowledge base it is bound to. There are three modes; you normally only pick between a first full load and the ongoing incremental sync — the cleanup runs alongside.

Full load¶

The first sync (and any manual re-index) is a full load: every supported module in every enrolled course is fetched and indexed. Use this to seed a new knowledge base, or to rebuild it after big changes.

Incremental sync¶

Routine syncs are incremental: the connector re-reads the course structure but only re-indexes modules whose timecreated / timemodified (or a file's modification time) falls inside the sync window. New and changed items are picked up; untouched items are left as they are. Because document ids are stable (moodle_<type>_<id>), a changed item replaces its existing document instead of duplicating it.

Removing deleted items (stale cleanup)¶

The data source has a Sync deleted files option; it is off by default. With it on, the connector produces alongside each sync a lightweight snapshot of the ids of every module that could be indexed (no downloads). RAGflow compares that list to what is in the knowledge base and deletes any indexed document whose Moodle module no longer exists (or is no longer reachable by the token user). So un-enrolling the connector's user from a course, or deleting an activity, removes the corresponding documents on the next sync.

What triggers a re-index of a single item¶

An item is re-indexed when its module or its file reports a newer timemodified. Editing a page or book, posting to a forum, replacing a resource file, or changing an activity's description all bump the timestamp and cause a refresh on the next incremental sync.

Practical notes¶

Scope follows enrolment. The set of indexed courses is exactly the token user's enrolled courses at sync time — enrol the user in a new course and it appears next sync; un-enrol and its documents are cleaned up.
Parsing is RAGflow's job. The connector delivers the raw documents; RAGflow then parses/embeds them into the knowledge base with that KB's configured pipeline and embedding model.
Frequency is set on the data source itself, not per knowledge base: the Refresh Freq field (in minutes) in the Moodle source's settings. Pick a cadence that matches how often your courses change — a short interval on a large site means syncs overlap or run into Timeout Secs.

See also: How it works · Troubleshooting.

Note: This documentation was created with the help of AI. Spotted an error, an omission, or something unclear? Please report it in the ragcon-docs GitHub repository so we can fix it.