Seam Graphile Worker

This is a wrapper for Seam's opinionated usage of Graphile Worker.

Usage

seam-graphile-worker is made for usage with server projects. It will automatically connect to a database, run migrations, configure crontabs etc.

1. Install via npm add seam-graphile-worker
2. Run seam-graphile worker init to create a start:worker script, a seam-graphile-worker.config.ts file, and some other boilerplate files in the src/worker directory and src/tasks directory.
3. To run your worker, run npm run start:worker

Features

• Automatically parse environment database connection params via pg-connection-from-env
• Workers log errors to Sentry
• Workers have a health endpoint
• See all your healthy workers by polling seam_graphile_worker.worker_health
• Automatically restart jobs that were taken by an unhealthy worker
• Task index is linted to include all tasks with predictable names
• Define your cron jobs in a simple typed file format
• Automatically validate input payloads against zod schemas
• Prevent business-logic errors from retrying automatically
• Run jobs or crons in tests easily

Defining A New Task

Tasks

// e.g. src/tasks/my_task.ts
import { withTaskSpec } from "worker/with-task-spec"

export default withTaskSpec({
  task_name: "my_task",
})(async (payload, opts) => {
  // ...
})

Customizing Worker Task Middleware

To customize worker middleware, edit the src/worker/with-task-spec.ts file.

import {
  withSentry,
  withDatabasePool,
  createWithTaskSpec,
} from "seam-graphile-worker"

export const withTaskSpec = createWithTaskSpec({
  middlewares: [withSentry, withDatabasePool],
})

Defining New Task Middleware

You can define new Task Middleware by creating files in src/worker/middlewares/with-*.ts

// e.g. src/worker/middlewares/with-workspace.ts
import { TaskMiddleware } from "seam-graphile-worker"

export const withWorkspace: TaskMiddleware<{
  opts_output: {
    workspace: { name: string; created_at: Date }
  }
  opts_dependencies: {
    db: DatabaseConnection
  }
  payload_dependencies: {
    workspace_id: number
  }
}> = (next) => async (payload, opts) => {
  opts.workspace = await opts.db.one("SELECT * FROM workspace WHERE id = $1", [
    payload.workspace_id,
  ])

  return next(payload, opts)
}

Defining Cron Jobs

To define cronjobs, edit crontabs property of seam-graphile-worker.config.ts

import * as tasks from "worker/tasks"

export default {
  tasks,
  crontabs: [
    {
      task: "example_task",
      frequency: "1h",
      payload: { some_param: "some_value" }
      options: { priority: 1 }
    }
  ]
}

Deploying to Fly

Technical Details

seam-graphile-worker maintains two schemas:

• graphile_worker - this is the graphile worker schema, this is not changed so that we have the same schema structure as upstream graphile worker
• seam_graphile_worker - this schema adds additional information to the context for graphile_worker

Stuff seam-graphile-worker doesn't do that you should do

• Introduce logging middleware that captures logs from your job, the payload, and other information then inserts it to a log table or sends to a logging platform.