Since TON nodes uses its own ADNL binary transport protocol, a intermediate service is needed for an HTTP connection.

TON HTTP API is such a intermediate service, receiving requests via HTTP, it accesses the lite servers of the TON network using tonlibjson.

You can use the ready-made toncenter.com service or start your own instance.

Recommended hardware:

• CPU architecture: x86_64 or arm64.
• HTTP API only: 1 vCPU, 2 GB RAM.
• HTTP API with cache enabled: 2 vCPUs, 4 GB RAM.

There are two main ways to run TON HTTP API:

• Local (experimental): works on following platforms: Ubuntu Linux (x86_64, arm64), MacOSX (Intel x86_64, Apple M1 arm64) and Windows (x86_64).
• Docker Compose: flexible configuration, recommended for production environments, works on any x86_64 and arm64 OS with Docker available.

Note: It is simple but not stable way to run the service. We do not recommend to use it in production.

• (Windows only, first time) Install OpenSSL v1.1.1 for win64 from here.
• Install package: pip install ton-http-api.
• Run service with ton-http-api. This command will run service with mainnet configuration.
  • Run ton-http-api --help to show parameters list.

• (First time) Install required tools: docker, docker-compose, curl.
  • For Ubuntu: run scripts/setup.sh from the root of the repo.
  • For MacOS and Windows: install Docker Desktop.
  • Note: we recommend to use Docker Compose V2.
• Download TON configuration files to private folder:

  mkdir private
  curl -sL https://ton-blockchain.github.io/global.config.json > private/mainnet.json
  curl -sL https://ton-blockchain.github.io/testnet-global.config.json > private/testnet.json

• Run ./configure.py to create .env file with necessary environment variables (see Configuration for details).
• Build services: docker-compose build.
  • Or pull latest images: docker-compose pull.
• Run services: docker-compose up -d.
• Stop services: docker-compose down.

You should specify environment parameters and run ./configure.py to create .env file. bash export TON_API_LITESERVER_CONFIG=private/testnet.json ./configure.py

The service supports the following environment variables:

• TON_API_HTTP_PORT (default: 80)

  Port for HTTP connections of API service.

• TON_API_ROOT_PATH (default: /)

  If you use a proxy server such as Nginx or Traefik you might change the default API path prefix (e.g. /api/v2). If so you have to pass the path prefix to the API service in this variable.

• TON_API_WEBSERVERS_WORKERS (default: 1)

  Number of webserver processes. If your server is under high load try increase this value to increase RPS. We recommend setting it to number of CPU cores / 2.

• TON_API_GET_METHODS_ENABLED (default: 1)

  Enables runGetMethod endpoint.

• TON_API_JSON_RPC_ENABLED (default: 1)

  Enables jsonRPC endpoint.

• TON_API_LOGS_JSONIFY (default: 0)

  Enables printing all logs in json format.

• TON_API_LOGS_LEVEL (default: ERROR)

  Defines log verbosity level. Values allowed: DEBUG,INFO,WARNING,ERROR,CRITICAL.

• TON_API_GUNICORN_FLAGS (default: empty)

  Additional Gunicorn command line arguments.

• TON_API_TONLIB_LITESERVER_CONFIG (default docker: private/mainnet.json local: https://ton.org/global-config.json)

  Path to config file with lite servers information. In case of native run you can pass URL to download config. Docker support only path to file.

• TON_API_TONLIB_KEYSTORE (default docker: /tmp/ton_keystore local: ./ton_keystore/)

  Path to tonlib keystore.

• TON_API_TONLIB_PARALLEL_REQUESTS_PER_LITESERVER (default: 50)

  Number of maximum parallel requests count per worker.

• TON_API_TONLIB_CDLL_PATH (default: empty)

  Path to tonlibjson binary. It could be useful if you want to run service on unsupported platform and have built the libtonlibjson library manually.

• TON_API_TONLIB_REQUEST_TIMEOUT (default: 10)

  Timeout for liteserver requests.

• TON_API_CACHE_ENABLED (default: 0)

  Enables caching lite server responses with Redis.

• TON_API_CACHE_REDIS_ENDPOINT (default: localhost, docker: cache_redis)

  Redis cache service host.

• TON_API_CACHE_REDIS_PORT (default: 6379)

  Redis cache service port.

• TON_API_CACHE_REDIS_TIMEOUT (default: 1)

  Redis cache timeout.

To point the HTTP API to your own lite server you should set TON_API_TONLIB_LITESERVER_CONFIG to config file with your only lite server.

• If you use MyTonCtrl on your node you can generate config file with these commands:

  $ mytonctrl
  MyTonCtrl> installer
  MyTonInstaller> clcf
  

  Config file will be saved at /usr/bin/ton/local.config.json.
• If you don't use MyTonCtrl: copy private/mainnet.json and overwrite section liteservers with your liteservers ip, port and public key. To get public key from liteserver.pub file use the following script:

  python -c 'import codecs; f=open("liteserver.pub", "rb "); pub=f.read()[4:]; print(str(codecs.encode(pub,"base64")).replace("\n",""))'
  

• Once config file is created assign variable TON_API_TONLIB_LITESERVER_CONFIG to its path, run ./configure.py and rebuild the project.

• Clone the repo as many times as many instances you need to the folders with different names (otherwise docker compose containers will conflict).
• Configure each instance to use unique port (env variable TON_API_HTTP_PORT)
• Build and run every instance.

Binary file libtonlibjson now moved to pytonlib.

• Docker Compose: docker-compose build --no-cache.
• Local run: pip install -U ton-http-api.

Usually, liteservers from the config has already deleted the block, which specified in init_block section. To update init block, please backup your config file and run script ./scripts/update_init_block.sh private/mainnet.json. For testnet add flag --testnet.