registry-backend/README.md

registry-backend

The server that backs Hanzo Registry, a public collection of custom node packs used in Hanzo Studio. Read the docs to publish your first node. Browse existing nodes at https://registry.hanzo.ai.

The backend API server for hanzo registry and Comfy CI/CD.

Discord

Registry Frontend

cli

Local Development

Golang

Install Golang here.

Install go packages

go get

Atlas

Atlas is used for editing the database, so install it from https://github.com/ariga/atlas/?tab=readme-ov-file#quick-installation

Supabase

Install Supabase Cli

brew install supabase/tap/supabase

supabase start

Open Supabase Studio locally.

Set up local ADC credentials

Install GCloud CLI https://cloud.google.com/sdk/docs/install

These are needed for authenticating Firebase JWT token auth + calling other GCP APIs.

When testing login with registry, use this: gcloud config set project dreamboothy-dev

gcloud auth application-default login

If you are testing creating a node, you need to impersonate a service account because it requires signing cloud storage urls.

gcloud auth application-default login --impersonate-service-account [email protected]

TODO(robinhuang): Create a service account suitable for dev.

Start API Server

docker compose up

This commands starts the server with Air that listens to changes. It connects to the Supabase running locally.

Code Generation

Make sure you install the golang packages locally.

go get

Schema Change

Update the files in ent/schema.

Regenerate code

This should search all directories and run go generate. This will run all the commands in the generate.go files in the repository.

go generate ./...

Or manually run:

go run -mod=mod entgo.io/ent/cmd/ent generate --feature sql/upsert --feature sql/lock --feature sql/modifier ./ent/schema

Generate Migration Files

Run this command to generate migration files needed for staging/prod database schema changes:

atlas migrate diff migration --dir "file://ent/migrate/migrations" --to "ent://ent/schema" --dev-url "docker://postgres/15/test?search_path=public"

API Spec Change (openapi.yml)

Regenerate code

This should search all directories and run go generate. This will run all the commands in the generate.go files in the repository.

go generate ./...

Or manually run:

export PATH="$PATH:$HOME/bin:$HOME/go/bin"

https://github.com/deepmap/oapi-codegen/issues/795

oapi-codegen --config drip/codegen.yaml openapi.yml

TroubleShooting / Common Errors

Here are some common errors and how to resolve them.

Security Scan

If you are calling the security-scan endpoint, you need to add the endpoint url to docker-compose.yml and then make sure you have the correct permissions to call that function.

Check the security-scan Cloud Function repo for instructions on how to do that with gcloud.

For non hanzoui contributors, you can use your own hosted function or just avoid touching this part. We keep the security scan code private to avoid exploiters taking advantage of it.

Firebase Token Errors

Usually in localdev, we use dreamboothy-dev Firebase project for authentication. This conflicts with our machine creation logic because all of those machine images are in dreamboothy. TODO(robinhuang): Figure out a solution for this. Either we replicate things in dreamboothy-dev, or we pass project information separately when creating machine images.

Creating VM instance error

Example:

{
    "severity": "ERROR",
    "error": "error creating instance: Post \"https://compute.googleapis.com/compute/v1/projects/dreamboothy/zones/us-central1-a/instances\": oauth2: \"invalid_grant\" \"reauth related error (invalid_rapt)\" \"https://support.google.com/a/answer/9368756\"",
    "time": "2024-02-26T01:32:27Z",
    "message": "Error creating instance:"
}

{
    "severity": "ERROR",
    "error": "failed to get session using author id 'nz0vAxfqWLSrqPcUhspyuOEp03z2': error creating instance: Post \"https://compute.googleapis.com/compute/v1/projects/dreamboothy/zones/us-central1-a/instances\": oauth2: \"invalid_grant\" \"reauth related error (invalid_rapt)\" \"https://support.google.com/a/answer/9368756\"",
    "time": "2024-02-26T01:32:27Z",
    "message": "Error occurred Path: /workflows/:id, Method: GET\n"
}

Resolution:

You would likely need to run gcloud auth application-default login again and restart your docker containers/services to pick up the new credentials.

Calling CreateSession endpoint

Use the postman collection to call the CreateSession endpoint. You should be able to import changes with openapi.yml file. You should use this as a request body since there are list of supported GPU type.

{
  "gpu-type": "nvidia-tesla-t4"
}

Bypass Authentication Error

In order to bypass authentication error, you can add make the following changes in firebase_auth.go file.

package drip_middleware

func FirebaseMiddleware(entClient *ent.Client) echo.MiddlewareFunc {
 return func(next echo.HandlerFunc) echo.HandlerFunc {
  return func(ctx echo.Context) error {
   userDetails := &UserDetails{
    ID:    "test-james-token-id",
    Email: "[email protected]",
    Name:  "James",
   }

   authdCtx := context.WithValue(ctx.Request().Context(), UserContextKey, userDetails)
   ctx.SetRequest(ctx.Request().WithContext(authdCtx))
   newUserError := db.UpsertUser(ctx.Request().Context(), entClient, userDetails.ID, userDetails.Email, userDetails.Name)
   if newUserError != nil {
    log.Ctx(ctx).Info().Ctx(ctx.Request().Context()).Err(newUserError).Msg("error User upserted successfully.")
   }
   return next(ctx)
  }
 }
}

Machine Image Not Found

We use a custom machine image to create VM instances. That machine image is specified in docker-compose.yml file.

MACHINE_IMAGE: "comfy-cloud-template-3"

If you are getting an error that the machine image is not found, you can create a new machine image by following the steps below:

TODO: explore steps to create machine image with comfy setup.

For the purpose of just testing endpoints, you don't really need to worry about Comfy specific machine image. You can simply create a new VM on the GCP console and use that VM's image to create a new machine image. And then update the docker-compose.yml file with the new machine image name.

Clean Up Resources

You can use this script to cleanup resources for specific user.

`docker compose -f scripts/cleanup/docker-compose.cleanup.yml run --rm cleanup -u <user id>`