@jamsocket/server

JavaScript/TypeScript library for spawning session backends server-side.

Read the docs here

Installation

npm install @jamsocket/server

Example

Here's an example of how different parts of Jamsocket's client libraries work together.

import { Jamsocket } from '@jamsocket/server'

const jamsocket = new Jamsocket({
   account: '[YOUR ACCOUNT]',
   token: '[YOUR TOKEN]',
   service: '[YOUR SERVICE]',
   // during development, you can simply pass { dev: true }
})

const connectResponse = await jamsocket.connect() // returns an instance of ConnectResponse

import {
  SessionBackendProvider, SocketIOProvider,
  useEventListener, useSend, useReady
} from '@jamsocket/socketio'

function Root() {
  return(
    <SessionBackendProvider connectResponse={connectResponse}>
      <SocketIOProvider url={connectResponse.url}>
        <MyComponent />
      </SocketIOProvider>
    </SessionBackendProvider>
  )
}

function MyComponent() {
  const ready = useReady()
  const sendEvent = useSend()

  useEffect(() => {
    if (ready) {
      sendEvent('some-event', someValue)
    }
  }, [ready])

  useEventListener('another-event', (args) => {
    // do something when receiving an event message from your session backend...
  })
}

Library Reference

@jamsocket/server

Jamsocket

Create a Jamsocket instance with the Jamsocket class from @jamsocket/server folder.

In local development, you can simply set dev to true.

import { Jamsocket } from '@jamsocket/server'
const jamsocket = new Jamsocket({ dev: true })

In production, provide your account, token, and service information.

import { Jamsocket } from '@jamsocket/server'
const jamsocket = new Jamsocket({
  account: '[YOUR ACCOUNT]',
  token: '[YOUR TOKEN]',
  service: '[YOUR SERVICE]',
})

fromEnv(env)

The Jamsocket class comes with a static method that returns an instance of Jamsocket configured by the provided environment. The fromEnv() method expects to find either JAMSOCKET_TOKEN, JAMSOCKET_ACCOUNT, and JAMSOCKET_SERVICE values or a JAMSOCKET_DEV: true value. If running in dev mode, the function will also accept an optional JAMSOCKET_DEV_PORT which tells the Jamsocket instance where to find the dev server. (This is only needed if you're running npx jamsocket dev with a custom port.)

Example:

import { Jamsocket } from '@jamsocket/server'
const jamsocket = Jamsocket.fromEnv(process.env)

connect()

The Jamsocket instance includes a connect function that you can use to get a connection URL for a session backend. If you provide a key, the connect function will either spawn a new backend or return the running backend that holds the provided key if one exists. When generating a connection URL for a backend, you can provide an optional ConnectRequest object. It returns a promise, which resolves with a ConnectResponse.

If the underlying connect request fails with a non-200 status code, the returned Promise will reject with an HTTPError.

Learn more about the various options you can pass in a ConnectRequest in our API docs.

A Jamsocket class should only be instantiated on the server since it takes a Jamsocket Auth Token which must be kept secret.

import { Jamsocket } from '@jamsocket/server'
const jamsocket = new Jamsocket({
  account: '[YOUR ACCOUNT]',
  token: '[YOUR TOKEN]',
  service: '[YOUR SERVICE]',
})

const connectResponse = await jamsocket.connect() // no options are required

// or

const connectResponse = await jamsocket.connect({
  key: 'my-key',
  spawn: {
    lifetime_limit_seconds: 432_000, // 10 hours
    max_idle_seconds: 300, // 5 minutes
    executable: {
      env: {
        'MY_ENV_VAR': 'foo'
      }
    }
  },
  user: 'my-user-id', // optional user identifier to be included in request headers to session backend
  auth: { 'my_user_metadata': 'bar' } // optional values to be JSON-serialized and included in request headers to session backend
})

status(backendId)

The Jamsocket instance includes a status function that you can use to get the current status of a backend. It takes a backend ID and returns a Promise<BackendState>.

If the underlying status request fails with a non-200 code, the returned Promise will reject with an HTTPError. This includes if no backend is found with the given ID.

Learn more about the various statuses backends can have in our API docs.

import { Jamsocket } from '@jamsocket/server'
const jamsocket = new Jamsocket({
  account: '[YOUR ACCOUNT]',
  token: '[YOUR TOKEN]',
  service: '[YOUR SERVICE]',
})

const backendState = await jamsocket.status(backendId)

statusStream(backendId, onStatusCallback)

The Jamsocket instance includes a statusStream function that you can use to stream a backend's status updates. It takes a backend ID and an onStatusCallback ((backendState: BackendState) => void), which will be called on each status as it is received. The statusStream() function returns a Promise<UnsubscribeFn>. The underlying stream is closed when either the backend reaches a terminated state or the UnsubscribeFn is called. If the underlying stream closes but the backend has not terminated and the UnsubscribeFn has not been called, this function will automatically reconnect.

If the underlying status request fails with a non-200 code, the returned Promise will reject with an HTTPError. This includes if no backend is found with the given ID.

The onStatusCallback function will be called exactly once for every status the backend has encountered, including statuses from before statusStream() was called.

Learn more about the various statuses backends can have in our API docs.

import { Jamsocket } from '@jamsocket/server'
const jamsocket = new Jamsocket({
  account: '[YOUR ACCOUNT]',
  token: '[YOUR TOKEN]',
  service: '[YOUR SERVICE]',
})

const unsubscribe = await jamsocket.statusStream(backendId, (backendState) => {
  console.log(backendState.status, backendState.time)
})

unsubscribe()

Types

type JamsocketInitOptions =
  | {
      account: string
      token: string
      service: string
      apiUrl?: string
    }
  | {
      dev: true
      port?: number
    }

type ConnectRequest = {
  key?: string
  spawn?:
    | boolean
    | {
        tag?: string
        cluster?: string
        lifetime_limit_seconds?: number
        max_idle_seconds?: number
        executable?: {
          mount?: string | boolean
          env?: Record<string, string>
          resource_limits?: {
            // The CPU period (in microseconds), defaults to 100_000 (100ms)
            cpu_period?: number
            // Proportion of period the container is allowed to use (in percent, e.g. 100 = 100%)
            cpu_period_percent?: number
            // Total cpu time allocated to container (in seconds)
            cpu_time_limit?: number
            memory_limit_bytes?: number
            disk_limit_bytes?: number
          }
        }
      }
  user?: string
  auth?: Record<string, any>
}

type ConnectResponse = {
  backend_id: string
  spawned: boolean
  status: BackendStatus
  token: string
  url: string
  secret_token?: string | null
  status_url: string
  ready_url: string
}

type BackendStatus =
  | 'scheduled'
  | 'loading'
  | 'starting'
  | 'waiting'
  | 'ready'
  | 'terminating'
  | 'hard-terminating'
  | 'terminated'

type TerminationKind = 'soft' | 'hard'
type TerminationReason =
  | 'swept'
  | 'external'
  | 'keyexpired'
  | 'lost'
  | 'startuptimeout'
  | 'internalerror'

type BackendState =
  | { status: 'scheduled'; time: string }
  | { status: 'loading'; time: string }
  | { status: 'starting'; time: string }
  | { status: 'waiting'; time: string }
  | { status: 'ready'; time: string }
  | { status: 'terminating'; time: string; termination_reason: TerminationReason }
  | { status: 'hard-terminating'; time: string; termination_reason: TerminationReason }
  | {
      status: 'terminated'
      time: string
      termination_reason?: TerminationReason | null
      termination_kind?: TerminationKind | null
      exit_error?: boolean | null
    }