govuk-chat: Populating OpenSearch
We use OpenSearch to store an index of GOV.UK content which has been chunked. In production environments we populate this data by consuming the Publishing API message queue. This document describes how the search index can be populated outside these environments.
In development
This repo contains a number of seed files that can be used to populate a small search index of chunked content. These are stored in db/chunked_content_seeds/ with a README to explain how they are generated.
These can be populated by running:
bundle exec rake search:populate_chunked_content_index_from_seeds
If you need to populate a development instance with actual GOV.UK content, for a more representative search experience there are steps documented.
In Heroku
In our Heroku pipeline we share a single instance of OpenSearch across all the GOV.UK chat application instances. This instance is hosted on AWS in the govuk-test environment and is tagged with chat-engine-test. The credentials for accessing OpenSearch are available in AWS Secret Manager, on govuk-integration, under: govuk/govuk-chat/opensearch-test.
This instance has it's data populated nightly with a copy of productions search index. So it should not need to be updated.
To work with this search index locally you should grab the read-only credentials from the secret (the ones suffixed with -ro) and set env vars of OPENSEARCH_URL=, OPENSEARCH_USERNAME= and OPENSEARCH_PASSWORD=.
You can use the other credentials, the master user, if we ever need to write to this index.