Skip to content

Latest commit

 

History

History
127 lines (98 loc) · 3.64 KB

dev_environment.md

File metadata and controls

127 lines (98 loc) · 3.64 KB

Local Backend Development Environment

This document describes how to setup a local development environment for Kai's backend.

Overview

Running Kai's backend involves running a few processes:

  • Postgres instance which we deliver via container
  • Backend REST API server
  • [Optional] Hub Importer process to sync data from Konveyor

Prerequisites

  1. Python 3.11 or 3.12
  2. Podman
  3. Access to a Large Language Model (Note that results vary widely between models.)

Note

If you want to run Kai against an LLM you will likely need to configure a LLM API key to access your service (unless running against a local model).

We do provide a means of running Kai against previously cached data from a few models to aid demo flows. This allows you to run through the steps of using Kai without requiring access to a LLM. We call this KAI__DEMO_MODE, i.e. KAI__DEMO_MODE=true make run-server

If you do not provide LLM API access then KAI__DEMO_MODE will only be able to replay previous cached responses.

Setup

First, clone the repo and ensure you have the virtual environment setup

git clone https://github.com/konveyor-ecosystem/kai.git
cd kai
python3 -m venv env
source env/bin/activate
pip install pip-tools
pip install -e .

Next, open a new terminal run the postgres container via podman:

make run-postgres

Finally, return to your previous terminal run the Kai server:

make run-server

Note

If you want to run with cached LLM responses run with: KAI__DEMO_MODE=true make run-server. The KAI__DEMO_MODE option will cache responses and play them back on subsequent runs.

If you want to run with debug information, set the environment variable with: KAI__LOG_LEVEL=debug make run-server.

On your first run, there will be no solved examples in the database. You can load some sample data to get started. Open a new terminal and run:

source env/bin/activate
pushd samples
./fetch_apps.py
popd
make load-data

This should complete in ~1-2 minutes.

Run an analysis of a sample app (example for MacOS)

Note

We have checked in analysis runs for all sample applications so you do NOT need to run analysis yourself. The instructions below are ONLY if you want to recreate, this is NOT required

Ensure you have the source code for the sample applications checked out locally:

cd ./samples
./fetch_apps.py

This will check out the sample app source code to: ./kai_solution_server/samples/sample_repos.

Next, run the analysis:

cd macos
# Sets up the podman VM on MacOS so it will mount the host filesystem into the VM
./restart_podman_machine.sh
# Fetches kantra (our analyzer tool) and stores it in ../bin
./get_latest_kantra_cli.sh
cd ..
# Analyzes all sample apps we know about, in both the 'initial' and 'solved'
# states, expect this to run for ~2-3 hours.
./analyze_apps.py

The analysis data will be stored in: kai_solution_server/samples/analysis_reports/{APP_NAME}/<initial|solved>/output.yaml

Build and test a local image

  1. cd to the top level kai checkout
  2. Build the local image: podman build -f build/Containerfile . -t quay.io/konveyor/kai:local
  3. Run: TAG="local" podman compose up
  4. Then proceed with testing as you want