@metmirr/hydra-indexer

Hydra Indexer

Hydra Indexer is a daemon that ingests raw substrate data such as events and extrinsics from a substrate chain and saves it into a Postgres database. It is typically used in conjunction with Hydra Indexer Gateway providing a GraphQL API interface.

Installation

The indexer exits if an unrecoverable error occurs. It is therefore advisable to run the indexer with a process manager, such as PM2 or Docker.

Prerequisites

• Node 12.x
• Postgres database
• Redis instances
• (Optional) Docker

Environment variables

The indexer is set up using the following environment variables

Variable	Default	Required	Description
WS_PROVIDER_ENDPOINT_URI	-	Yes	Substrate API endpoint to connect
REDIS_URI	-	Yes	Redis instance URI
DB_NAME	-	Yes	Database name
DB_PORT	-	Yes	Database port
DB_HOST	-	Yes	Database host
DB_USER	-	Yes	Database user
DB_PASS	-	Yes	Database password
TYPES_JSON	-	No	Path to a JSON type definition with custom Substrate types
BLOCK_HEIGHT	0	No	Block height to start indexing. Ignored if the database already contains indexed blocks

Manual setup

Run

yarn && yarn build

For setting up the database and running the migrations, run yarn db:bootstrap. Make sure the environment variables DB_* are set.

For starting the indexer, run yarn start:prod.

Docker image

There are pre-built runnable docker images in joystream/hydra-indexer docker repo. The default command is yarn start:prod.

First, bootstrap the database:

docker run -e DB_HOST=... -e DB_PORT=... -e DB_NAME=... -e DB_PASS=... -e DB_USER=... joystream/hydra-indexer sh -c 'yarn db:bootstrap'

Then run the indexer (make sure that all the required environment variables are set)

docker run -e ... joystream/hydra-indexer

Advanced environment variables

Some optional environment variables are available for fine-tuning.

Variable	Default	Description
BLOCK_CACHE_TTL_SEC	60*60	TTL for processed blocks in the Redis cache
INDEXER_HEAD_TTL_SEC	60*15	TTL for the indexer head block entry
WORKERS_NUMBER	5	Number of concurrent workers fetching the blocks
BLOCK_PRODUCER_FETCH_RETRIES	3	Number of attempts fetching each a block before throwing an error. Set to -1 for indefinite attempts
SUBSTRATE_API_TIMEOUT	1000 * 60 * 5	Timeout in (milliseconds) for API calls
NEW_BLOCK_TIMEOUT_MS	60 * 10 * 1000	Panic if no blockchain blocks have been received within this time