Elastic App Search Knowledge Base Demo

Build and run frontend for testing

Frontend only

If you wish to test the frontend only, you can build and run the frontend docker image, you can run the following command from the project root directory:

docker build . --target=frontend -t eas-kb-demo/frontend
docker run -p5000:5000 --rm --env-file=eas-kb-demo-frontend/.env eas-kb-demo/frontend

You can now use the frontend from your browser at http://host.docker.internal:5000.

Note:

The docker build is a snapshot version of the production build to be deployed.
Data from an ESS App Search instance
If you want to develop the frontend, read the development documentation bellow.

Full stack

If you wish to test the full stack build locally, you have to boot all the required services (Elasticsearch, App Search) using the docker-compose.yml config with the project.

docker-compose up -d

It could take few minutes, for the stack to be fully booted. Once finished, you should be able to access:

App Search admin at http://host.docker.internal:3002 using enterprise_search/ password credentials
The frontend at http://host.docker.internal:5000

Note:

All data are automatically imported when starting the stack. It can take few minutes before the result start to appear in the frontend.
You can tune the engine relevance, curations, and synonyms by using the helpdesk meta engine in App Search.
The docker-compose stack is not intended to develop but to test built images. Images need to be refreshed everytime a change is made to the code using docker-compose build.

If something goes wrong with the import (use docker-compose logs -f dataimport to view logs), you can restart it with:

docker-compose run dataimport start-dataimport.sh

To reset the whole stack (all data will be lost), use:

docker-compose down

Development

Install dependencies

The project can be installed by running:

yarn install

Frontend development

All sources of the frontend are located into the eas-kb-demo-frontend workspace of the main yarn project.

By default, the frontend is configured to use the production App Search instance deployed in Elastic Cloud. Because frontend is a R/O application it should not be an issue.

To start the frontend, you can use:

yarn start

To run cypress tests for the frontend, you can use:

yarn cypress run

When working on the import, you should use your own version of App Search during the development. The easiest way to get the stack up and running is to use the docker-compose file bundled with this demo:

docker-compose up -d

After the stack will have finished to start, you can then access to App Search through your browser at http://localhost:3002 with the following credentials:

login: enterprise_search
password: password

Notes :

This demo uses meta engine which are available only when using a Platinum license.
You have to start a new trial for your App Search instance using the following command:

docker-compose exec elasticsearch  curl -XPOST -uelastic:elasticpassword "localhost:9200/_license/start_trial?acknowledge=true"
docker-compose restart enterprise_search

When using Docker for Mac, please make sure you have at least 4GiB of memory allocated to Docker (Preferences > Advanced).
Out of memory errors can cause ElasticSearch or App Search container to be killed (docker logs --tail can help to detect it).

Importing the data

The demo uses two different data sources:

data scrapped from the Elastic documentation (dataimport/data/data/elastic-co-docs.json)
data scrapped from the Elastic discussion forum (dataimport/data/data/elastic-co-discuss.json)

You can import theses data into App Search by running:

yarn dataimport

Notes:

The script will create one search engine in App Search for each dataset.
The script will create one meta engine helpdesk that will be used to query all the dataset. If needed, relevance tunning should be done in the meta engine.
You will need to information during the setup:
- App Search API base URL: default is set to http://localhost:3002/api/as/v1/ (should be fine)
- App Search API private key: you can find it into the credentials section of the App Search dashboard (http://localhost:3002/as#/credentials)

anborg/app-search-kb-demo