Pragence

Confluence

Index the pages of the Confluence Cloud spaces you name. Connector type: confluence.

This connector follows the published Confluence Cloud REST API. It has not yet been verified against a live Confluence site. Run a first sync on one small space and check the run before relying on it.

Setup

  1. The deployment needs its own Atlassian OAuth app, with SOURCE_ATLASSIAN_CLIENT_ID and SOURCE_ATLASSIAN_CLIENT_SECRET set.
  2. In the console, choose Confluence under Add source and then Sign in with Atlassian. Through the API, call POST /api/v1/connections/atlassian/authorize, send the operator's browser to the returned address, and keep the connection_id that comes back. See Sign-in connections.
  3. Create the confluence connector with that sign-in and the settings below, then start a sync.
SettingDescription
space_keysRequired. The keys of the spaces to sync, for example ADM, HANDBOOK
site_urlThe site address, for example https://acme.atlassian.net. Needed only when the sign-in can read more than one site

What gets indexed

Each page and blog post in the chosen spaces becomes one document. Empty pages are left out. Comments and attachments are not indexed.

Access

Page restrictions are handled one of two ways, set by Respect page restrictions (mirror_permissions, admins only):

  • Off (default). A page with view restrictions (limited to named people or groups), either its own or inherited from any parent page, is left out and reported on the run as restricted_at_source. If a page becomes restricted later, the next sync deletes it from the index.
  • On. A restricted page is indexed for exactly the people and groups it names: a person by email address, when Atlassian shares it, and a group as g:<group name>, whose members an admin lists under the source's access groups. A person whose email Atlassian withholds gets no access (it fails closed). A page under restricted parent pages is available only to those named at every restricted level; a parent the sign-in cannot read grants nobody. Access is re-read on every sync and refreshed on its own every six hours.

Space permissions are not mirrored in either mode: an unrestricted page is available to everyone who can use the project, so choose spaces meant for that audience.

Sync behaviour

  • Every sync lists the chosen spaces in full, so removed pages are deleted from the index. Unchanged pages are skipped.
  • A refused sign-in fails the run with source_refused.
  • When Atlassian asks the connector to slow down, it waits for the time Atlassian gives and tries again.
  • Without site_url, a sign-in that can read more than one Confluence site fails with not_configured.

On this page