Backend service

Rafiki’s backend service is the software’s main service for handling business logic and external communication. The service is responsible for:

• Exposing the endpoints of the Open Payments APIs for public clients to initiate payments and look up information. This is the external communication piece.
• Exposing an Interledger connector to send and receive STREAM packets with peers.
• Enabling you to manage accounts and track liquidity.
• Exposing an internal GraphQL Backend Admin API for service operators to manage accounts, liquidity, and application settings, like peering relationships.

Requirements

The following are required when using the backend service.

• A Redis database as a cache to share STREAM connection details and total amounts received across processes
• A Postgres database, separate from the auth service’s database, for Open Payments resources
• TigerBeetle or a separate Postgres database for accounting liquidity

  Note

  You can use the same database instance of Postgres for the backend service, auth service, and accounting liquidities (if not using TigerBeetle). However, separate database schemas are required within the Rafiki instance to maintain boundaries between the objects being managed.

You must also set the environment variables for the backend service.

Open Payments

The backend service exposes the Open Payments APIs, which are auth-protected using an opinionated version of the Grant Negotiation and Authorization Protocol (GNAP). Review the auth service page for more details about grant authorization and authentication.

The backend service allows you to manage Open Payments incoming payments, quotes, and outgoing payments. Quotes and outgoing payments call the ILP connector, described in the next section, to send ILP packets. Quoting sends unfulfillable probe packets, for example to determine a transaction’s cost before executing the payment. Outgoing payments send packets as part of executing the payment.

Interledger connector

The backend service exposes an ILP connector to send and receive ILP packets between peers.

Some of the responsibilities of a connector include:

• Authenticating packets against ILP account credentials.
• Forwarding packets to a sender or receiver.
• Rejecting a packet for any number of reasons, including expiration, insufficient liquidity, rate limit exceeded, or if the amount exceeds the maxPacketAmount agreed to by the connector and its peer.
• Converting currencies.
• Fulfilling packets with an internal STREAM server.

The backend service includes an HTTP server listening on the configured CONNECTOR_PORT. Your connector can receive incoming packets via HTTP and/or direct calls from within the backend service. An incoming packet is only accepted if it’s from a configured peer and accompanied by your peer’s incoming HTTP authToken.

Similarly, if a packet’s destination address corresponds to a peer, your connector will forward the packet to your peer over HTTP, along with your peer’s outgoing HTTP authToken.

Auth tokens

You and your peer should have exchanged incoming and outgoing auth tokens as part of establishing your peering relationship.

A packet can either continue on to your peer via HTTP or terminate at your Rafiki instance’s STREAM server. If the packet terminates at your STREAM server, your connector will attempt to extract and decode the payment tag from the packet’s destination address. When your connector successfully matches the tag with a locally managed wallet address or incoming payment, the connector will not forward the packet. Instead, it will credit the corresponding balance and track the total amount received in Redis to support STREAM receipts. Packets addressed to a wallet address happen via SPSP.

GraphQL Backend Admin API

The backend service exposes a GraphQL Backend Admin API to manage assets, peers, wallet addresses, Open Payments resources, and several types of liquidity.

Environment variables

Variable	Helm Value Name	Default	Description
ADMIN_PORT	backend.port.admin	3001	Admin API GraphQL Server port
AUTH_SERVER_GRANT_URL		undefined	endpoint on the Open Payments Auth Server to request a grant
AUTH_SERVER_INTROSPECTION_URL		undefined	endpoint on the Open Payments Auth Server to introspect an access token
AUTO_PEERING_SERVER_PORT		3005	If auto-peering is enabled, its server will use this port
CONNECTOR_PORT	backend.port.connector	3002	port of the ILP connector for for sending packets over ILP over HTTP
DATABASE_URL	backend.postgresql.host, backend.postgresql.port, backend.postgresql.username, backend.postgresql.database, backend.postgresql.password	postgresql://postgres:password@localhost:5432/development	Postgres database URL of database storing the resource data; For Helm, these components are provided individually.
ENABLE_AUTO_PEERING		false	Flag to enable auto peering. View documentation.
ENABLE_SPSP_PAYMENT_POINTERS		true	enables SPSP route
EXCHANGE_RATES_LIFETIME	backend.lifetime.exchangeRate	15_000	time in milliseconds the exchange rates provided by the ASE via the EXCHANGE_RATES_URL are valid for
EXCHANGE_RATES_URL	backend.serviceUrls.EXCHANGE_RATES_URL	undefined	endpoint on the Account Servicing Entity to request exchange rates
GRAPHQL_IDEMPOTENCY_KEY_TTL_MS	backend.idempotencyTTL	86400000	TTL in milliseconds for idempotencyKey on GraphQL mutations (Admin API). Default: 24hrs
GRAPHQL_IDEMPOTENCY_KEY_LOCK_MS		2000	TTL in milliseconds for idempotencyKey concurrency lock on GraphQL mutations (Admin API)
ILP_ADDRESS	backend.ilp.address	undefined	ILP address of this Rafiki instance
ILP_CONNECTOR_URL		undefined	The ILP connector address where ILP packets are received. Communicated during auto-peering
INCOMING_PAYMENT_EXPIRY_MAX_MS		2592000000	Maximum milliseconds into the future incoming payments expiry can be set to on creation. Default: 30 days
INCOMING_PAYMENT_WORKERS	backend.workers.incomingPayment	1	number of workers processing incoming payment requests
INCOMING_PAYMENT_WORKER_IDLE	backend.workerIdle	200	time in milliseconds that INCOMING_PAYMENT_WORKERS will wait until they check an empty incoming payment request queue again
INSTANCE_NAME		undefined	this Rafiki instance’s name used to communicate for auto-peering
KEY_ID	backend.key.id	undefined	this Rafiki instance’s client key id
LOG_LEVEL	backend.logLevel	info	Pino Log Level
MAX_OUTGOING_PAYMENT_RETRY_ATTEMPTS		5	Maximum number of retry attempts for an outgoing payment before it is considered failed
NODE_ENV	backend.nodeEnv	development	node environment, development, test, or production
OPEN_PAYMENTS_PORT	backend.port.openPayments	3003	port of the Open Payments resource server port
OPEN_PAYMENTS_URL	backend.serviceUrls.OPEN_PAYMENTS_URL	undefined	public endpoint of this Open Payments Resource Server
OUTGOING_PAYMENT_WORKERS	backend.workers.outgoingPayment	4	number of workers processing outgoing payment requests
OUTGOING_PAYMENT_WORKER_IDLE	backend.workerIdle	200	time in milliseconds that OUTGOING_PAYMENT_WORKERS will wait until they check an empty outgoing payment request queue again
PRIVATE_KEY_FILE	backend.key.file	undefined	the path to this Rafiki instance’s client private key
QUOTE_LIFESPAN	backend.lifetime.quote	5 * 60_000	time in milliseconds an Open Payments quote is valid for
REDIS_TLS_CA_FILE_PATH	backend.redis.tlsCaFile	''	Redis TLS config
REDIS_TLS_CERT_FILE_PATH	backend.redis.tlsCertFile	''	Redis TLS config
REDIS_TLS_KEY_FILE_PATH	backend.redis.tlsKeyFile	''	Redis TLS config
REDIS_URL	backend.redis.host, backend.redis.port	redis://127.0.0.1:6379	Redis URL of database handling ILP packet data URL; For Helm, these components are provided individually.
SIGNATURE_SECRET	backend.quoteSignatureSecret	undefined	secret to generate request header signatures for webhook event requests
SIGNATURE_VERSION		1	version number to generate request header signatures for webhook event requests
SLIPPAGE	backend.ilp.slippage	0.01	accepted ILP rate fluctuation, default 1%
STREAM_SECRET	backend.ilp.streamSecret	undefined	seed secret to generate shared STREAM secrets
TIGERBEETLE_CLUSTER_ID		0	Tigerbeetle cluster ID picked by the system that starts the TigerBeetle cluster to create a Tigerbeetle client
TIGERBEETLE_REPLICA_ADDRESSES		3004	Tigerbeetle replica addresses for all replicas in the cluster, which are comma separated IP addresses/ports, to create a Tigerbeetle client
TRUST_PROXY		false	flag to use X-Forwarded-Proto header to determine if connections is secure
USE_TIGERBEETLE		true	flag - use TigerBeetle or Postgres for accounting
WALLET_ADDRESS_DEACTIVATION_PAYMENT_GRACE_PERIOD_MS		86400000	time in milliseconds into the future to set expiration of open incoming payments when deactivating wallet address. Default: 1 days
WALLET_ADDRESS_LOOKUP_TIMEOUT_MS		1500	time in milliseconds the ASE has to create a missing wallet address until timeout
WALLET_ADDRESS_POLLING_FREQUENCY_MS		100	frequency of polling while waiting for ASE to create a missing wallet address
WALLET_ADDRESS_URL	backend.serviceUrls.WALLET_ADDRESS_URL	http://127.0.0.1:3001/.well-known/pay	this Rafiki instance’s internal wallet address
WALLET_ADDRESS_WORKERS	backend.workers.walletAddress	1	number of workers processing wallet address requests
WALLET_ADDRESS_WORKER_IDLE	backend.workerIdle	200	time in milliseconds that WALLET_ADDRESS_WORKERS will wait until they check an empty wallet address request queue again
WEBHOOK_MAX_RETRY	backend.webhookMaxRetry	10	maximum number of times Rafiki backend retries sending a certain webhook event to the configured WEBHOOK_URL
WEBHOOK_TIMEOUT	backend.lifetime.webhook	2000	milliseconds
WEBHOOK_URL	backend.serviceUrls.WEBHOOK_URL	undefined	endpoint on the Account Servicing Entity that consumes webhook events
WEBHOOK_WORKERS	backend.workers.webhook	1	number of workers processing webhook events
WEBHOOK_WORKER_IDLE	backend.workerIdle	200	time in milliseconds that WEBHOOK_WORKERS will wait until they check an empty webhook event queue again
WITHDRAWAL_THROTTLE_DELAY	backend.withdrawalThrottleDelay	undefined	delay in liquidity withdrawal processing
ENABLE_MANUAL_MIGRATIONS	backend.enableManualMigrations	false	When set to true, user needs to run database manually with command npm run knex -- migrate:latest --env production