Content that doesn't show up correctly in search or list pages

Rummager (the search API) often gets out of sync with publishing applications. This affects any part of the site using it, including navigation pages and related links.

Root cause

Rummager depends on requests from publishing apps to stay up to date. This is usually a “fire and forget” task that doesn’t block the user’s publishing action, and if anything goes wrong, the search data will stay unchanged until the next time an update is made to the document.

The search team are working on improving this.

Check if search is the problem

A page with URL /council-tax can be queried using /api/search.json?filter_link=/council-tax. You can quickly switch between the two using the GOV.UK chrome plugin.

You can compare the data returned with the publishing app to check if it’s up to date. An empty response means search has never received the content.

You can also request different fields, for example /api/search.json?filter_link=/council-tax&fields=format,content_id.

Correct the search data

Most search issues can be fixed by either republishing the content, or manually reindexing via a rake task in the publishing app.

Published content is missing from search/finders/navigation

Try republishing the content.

Unpublished content is still showing up in search

Unpublished content can be removed from search manually using search admin.

Content is duplicated in search results

This has happened before when either the elasticsearch document type or id have changed.

You’ll need to look at the raw elasticsearch documents to see what happened.

There is a rake task to remove unwanted duplicates.

There are blank options on finder pages

On finder pages, facets on the left hand side can be populated from “link” fields in the search documents, like people, organisations, topics.

If the pages for those things don’t themselves exist in search, finder frontend won’t be able to fill in the titles.

This can be fixed by republishing/indexing those pages.