Skip to content

Latest commit

 

History

History
82 lines (49 loc) · 5.18 KB

DOCKER.md

File metadata and controls

82 lines (49 loc) · 5.18 KB
Raw

Using Vectara Answer with Docker

Introduction

Docker provides a container-based mechanism that ensures consistency and portability as well as environment isolation. For these reasons, deployment of Vectara Answer via Docker may be preferred in some cases.
To demonstrate how to use Vectara Answer with Docker we will now walk through the steps for building an application with the content of Paul Graham's essays website (http://paulgraham.com/).

We call this application AskPG.

Pre-Requisites

For Docker, you will first need to follow some additional installation steps:

Important

For mac users, please ensure you Docker Desktop is version 4.26.1 or above.

  1. Install Docker.
  2. Install pyyaml: pip3 install pyyaml.

Then, you will need to create the corpus:

  1. Log into the Vectara Console and create a corpus (https://docs.vectara.com/docs/console-ui/creating-a-corpus).
  2. Use Vectara Ingest to crawl Paul Graham's website and index it into your data store. Of course, you can also index this data on your own via the API.

Run Docker Desktop

Find and open the Docker Desktop application. When you run your application, it will use Docker to host your application inside a container.

Setup secrets.toml

Duplicate the secrets.example.toml file and rename the copy to secrets.toml.
Edit the secrets.toml file and change the api_key value under "default" to be your Vectara API key.

Note

The use of secrets.toml allows you to create multiple profiles, and store a differnet API key under each profile. This may be helpful if you are working with multiple Vectara Answer apps in parallel. As you can see in the included secrets.example.toml file we've included the API keys (query only) for the existing datasets that are part of the Quickstart: Feynman, vectara.com and vectara docs.

Note

Some secrets are shared among all profiles, and in order to avoid having to copy that secret under each profile, the code designates a special profile called [general] — each secret under this profile will be used in any profile (in addition to the profile-specific secrets). For example if you want to add a call to a Huggingface model you can include your hf_token there, or similarly if you want to call OpenAI or Anthropic you can include your key to their API.

Configuring your application

Make a duplicate of the config/vectara-website-search/ directory and rename it config/pg-search/. Update the config/pg-search/config.yaml file with these changes:

  • Change the corpus_id value to the ID of the corpus into which you ingested Paul Graham's essays as part of the vectara-ingest Quickstart.
  • Change the account_id value to the ID of your account. You can click on your username in the top-right corner to copy it to your clipboard.
  • Change the app_title to "Ask Paul Graham".

Edit the config/pg-search/queries.json file and update the four questions to a set of curated questions you'd like to include in the user interface. For example, "What is a maker schedule?"

4. Run the application

Execute the run script from the Vectara Answer root folder, specifying your configuration as config/pg-search, and selecting the default profile from your secrets file:

bash docker/run.sh config/pg-search default

The run.sh script performs the following steps:

  1. Runs the prepare_config.py python script which creates the .env file using the config.yaml and secrets.toml files
  2. Creates the Docker container and runs this container in your local Docker Desktop instance. The queries.json file is loaded dynamically as a Docker volume file.
  3. Exposes the app that runs inside the container, under localhost:80
  4. Waits 5 seconds (for the docker container to initialize), and open up a browser at localhost:80.

5. Done!

Your application is now up and running in Docker. Try a few queries to see how it works.

You can view your application's logs by running docker logs -f vanswer. You can stop the application and shut down the container with docker stop vanswer.

Customizing your application

With the Docker setup, the config.yaml file includes all of your configuration parameters as describe in README, and you can modify those as needed. In this case the parameters do not require the REACT_APP prefix (as in the case of running locall with the .env file described in the main README) - see examples in any of the configuration folders that are available in this repo.

Deploying Your App

With Docker, you can deploy your Vectara Answer application on a cloud service like Render by following these detailed instructions. Deployment on other cloud platforms that support docker like AWS, Azure or GCP is also possible.

FAQs

Why do I get a "Cannot connect to the Docker daemon" error when I try to run my application?

Docker needs to be running before you can run your application. Find and open the Docker app.