Skip to content

Latest commit

 

History

History
116 lines (74 loc) · 4.65 KB

custom_apps.md

File metadata and controls

116 lines (74 loc) · 4.65 KB
Raw

Run Kai with your Applications

Konveyor AI (KAI) leverages LLMs to streamline the modernization of application source code. To learn how Kai works, try this demo that migrates a JavaEE app to Quarkus.

Pre-requisites

  • Install podman
  • VS Code
  • git
  • GEN AI Credentials

To run Kai with your Applications, follow these steps

Configure Application

  1. Navigate to the samples folder - cd samples
  2. In ./kai_solution_server/samples/config.py, add your application's info to repos dict.
 "<app name>": [
        "https://github.com/<path to repo>",
        "<source branch>",
        "<migrated/target branch>",
    ],

  • app name : The name of your application.

  • path to repo : The full path to your application's repository on GitHub.

  • source branch : The branch of the repository that contains the current version of your application code.

  • migrated/target branch : The branch that will contain the migrated version of your application code. This is optional and will be used to generate solved examples that can be used to solve similar issues in the future.

Here's an example for an application named exampleApp:

"exampleApp": [
        "https://github.com/exampleOrg/exampleApp.git",
        "develop",  # Source branch
        "migrated",  # Target branch
    ],

Add your app information to sample_apps dict in the same file

"<app_name>" : "sample_repos/<app_name>",

for exampleApp, it will be of the format

    "exampleApp" = "sample_repos/exampleApp"

Make sure that the git repo is public and accessible. Currently, Kai only supports publicly accessible repos

Fetch the Application

Once you are done with the above configuration, run

./fetch_apps.py

This step will clone the repo in the location 'sample_repos'

Run Analysis

Before running analysis, make sure that you have kantra installed

  1. cd macos
  2. ./restart_podman_machine.sh # setups the podman VM on MacOS so it will mount the host filesystem into the VM
  3. ./get_latest_kantra_cli.sh # fetches 'kantra' our analyzer tool and stores it in ../bin
  4. cd ..

To ensure the analysis uses the necessary custom rules, follow these steps:

  • Download or create your custom rules and place them in the kai_solution_server/samples/custom_rules directory.

To run analysis reports for the applications listed in ./kai_solution_server/samples/config.py, run

./analyze_apps.py

This step analyzes all sample apps we know about, in both the 'initial' and 'solved' states, expect this to run for ~2-3 hours.

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

Run Kai server

  1. Configure GEN AI key and select model by following these steps
  2. Run Kai server by following these steps

Configure Kai VSCode plugin

  1. Install Kai IDE plugin by following these instructions

  2. Clone the application you want to analyze and navigate to File > Open in VSCode and locate the folder and open it

  3. Run Kantra analysis of the app after selecting the targets you need

  4. To fix issues identified during analysis using Kai, follow these steps

Note: There are two different places where analysis is performed and currently only one of them is set to ignore the default Kantra rules:

  1. Analysis of Sample Data: This runs on several sample applications using both default and custom rules. It simulates integration with a Konveyor instance populated with data. Configuration for this analysis is in analyze_apps.py.

  2. Client-Side Analysis: This can be handled through example/run_demo.py or running in an IDE. Default rules are ignored, and only custom rules are used. Configuration for this analysis is in analyze.sh

For client-side analysis, use the --enable-default-rulesets=false option and then pass in the below directory to use: Konveyor Default Rules