Clone project routing

Cloning an machine learning anomaly detection jobs copies the source job's datafeeds configuration, including project_routing. If you clone a job that searches linked projects, the new job inherits the same cross-project scope unless you change it during clone.

Symptoms

• A cloned job searches more linked projects than you expected.
• The new job's datafeeds shows the same project_routing as the source job in GET _ml/datafeeds/{datafeed_id}.
• Detection results or extraction timing on the clone match the source even though you intended a narrower scope.

Check the clone configuration

After creating the clone, expand the job row in the anomaly detection jobs list or call:

				GET _ml/datafeeds/{new_datafeed_id}
		

Compare project_routing on the clone with the source datafeeds. If they match and you did not edit scope in the clone dialog, the clone inherited the source routing.

Review search scope in the clone dialog

When cloning a job in Kibana, open the Search scope section before confirming create. The default shows the source project_routing. Edit it if the clone should search fewer projects — for example, _origin for local-only analysis.

Edit routing before creating the clone

Set the intended project_routing in the clone wizard rather than updating after create. Changing scope later on a running job may require closing the job and retaining a model snapshot.

Verify after create

Expand the new job in the anomaly detection jobs list and confirm the displayed search scope matches your intent. Optionally confirm with:

				GET _ml/datafeeds/{new_datafeed_id}/_stats
		

If the clone is too broad, see Scope too broad for how to narrow project_routing.