seer

Using Seer

Autofix

Codebase Index Storage

Autofix creates and uses a ChromaDB for every separate codebase. This can either be stored in a Google Cloud Storage (GCS) bucket, in your local filesystem, or you can make your own storage adapter.

Local Workspace

Autofix will need a local folder to use as its workspace, by default it uses data/chroma/workspaces.

You can set the location in filesystem where Autofix will use as a workspace with CODEBASE_WORKSPACE_DIR.

Google Cloud Storage

To use GCS, you need to set the following environment variables:

GOOGLE_APPLICATION_CREDENTIALS=<path to your GCS credentials file>
GOOGLE_CLOUD_PROJECT=<your GCS project ID>

Then, you can define your storage bucket with:

CODEBASE_STORAGE_TYPE=gcs
CODEBASE_GCS_STORAGE_BUCKET=<your GCS bucket name>
CODEBASE_GCS_STORAGE_DIR=<path within gcs bucket>

Local Filesystem

To use local filesystem, you can set the following environment variable:

CODEBASE_STORAGE_TYPE=filesystem
CODEBASE_STORAGE_DIR=<path to your local directory>

Local Development

Setup

Use direnv or a similar tool that sources .envrc. It will check your python version and setup virtualenv for you:

direnv allow

Recommended to use pyenv or similar python environment manager so as to be able to use differing python versions between sentry projects.

You will need pyenv installed and configured for python 3.11 before running direnv allow.

Environment variables

Set the environment variables that are in .env.example with the actual values, save this as .env in the root of the project.

The example shows GITHUB_PRIVATE_KEY and GITHUB_APP_ID. You can also use just GITHUB_TOKEN instead.

Add export SENTRY_AUTH_TOKEN=<your sentry auth token> to your rc file.

Install GCloud CLI

Refer https://cloud.google.com/sdk/docs/install. We use the super-big-data project.

Model Artifacts

You will need model artifacts to run inference in seer, get them from gcs by:

gsutil cp -r gs://sentry-ml/seer/models ./models

Running

To run for development locally in one ago including building the docker image and rabbitmq container:

make dev # runs docker compose up --build

Port 9091 will be exposed which is what the local sentry application will look for to connect to the service.

Database Stuff

Applying Migrations

make update

This will apply all migrations to the database and create new image.

Creating Migrations

make migration

Running Tests

make test

Opening a shell

make shell

Resetting Dev Environment

If you run into any data issue or connection issue and want to start from scratch, run the following set of commands from the command shell inside the seer repo:

docker compose down --volumes
make update && make dev