NEAR Lake is an indexer built on top of NEAR Indexer microframework to watch the network and store all the events as JSON files on AWS S3.

We used to have NEAR Indexer for Explorer that was watching for the network and stored all the events to PostgreSQL database. PostgreSQL became the main bottleneck for us. After some brainstorming sessions and researches we decided to go with AWS Aurora database.

Knowing the fact that NEAR Explorer is not the only project that uses the Indexer for Explorer's database, we wanted to come up with the concept that will allow us to cover even more projects that can benefit from the data from NEAR Protocol.

That's why we decided to store the data from the blockchain as JSON files on AWS S3 bucket that can be used as a data source for different projects.

As "Indexer for Explorer Remake" project we are going to have near-lake as a data writer. There's going to be another project that will read from AWS S3 bucket and will store all the data in SingleStore database. This will replace NEAR Indexer for Explorer PostgreSQL database at some moment and will become the main source for NEAR Explorer.

The final setup consists of the following components:

• AWS S3 Bucket as a storage
• NEAR Lake binary that operates as a regular NEAR Protocol peer-to-peer node, so you will operate it as any other Regular/RPC Node in NEAR

Before you proceed, make sure you have the following software installed:

• Rust compiler of the version that is mentioned in rust-toolchain file in the root of nearcore project.

• Ensure you have AWS Credentials configured From AWS Docs:

  For example, the files generated by the AWS CLI for a default profile configured with aws configure looks similar to the following.

  ~/.aws/credentials

  [default]
  aws_access_key_id=AKIAIOSFODNN7EXAMPLE
  aws_secret_access_key=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY
  

$ cargo build --release

To connect NEAR Lake to the specific chain you need to have necessary configs, you can generate it as follows:

$ ./target/release/near-lake --home ~/.near/testnet init --chain-id testnet --download-config --download-genesis

The above code will download the official genesis config and generate necessary configs. You can replace testnet in the command above to different network ID (mainnet).

NB! According to changes in nearcore config generation we don't fill all the necessary fields in the config file. While this issue is open near/nearcore#3156 you need to download config you want and replace the generated one manually.

• testnet config.json
• mainnet config.json

Configs for the specified network are in the --home provided folder. The configuration in the config.json file ensures that NEAR Lake follows all the necessary shards, so "tracked_shards" parameters in ~/.near/testnet/config.json is configured as follows: (nearcore treats empty value for "tracked_shards" as "do not track any shard" and any value as "track all shards".)

...
"tracked_shards": [0],
...

You can also configure NEAR Lake via environment variables. This is useful if you want to run NEAR Lake in a Docker container.

Add .env file to the root of the project with the following content:

BUCKET=near-lake-custom # name of the bucket to store data in (e.g. near-lake-data-testnet or near-lake-data-mainnet)
REGION=eu-central-1
AWS_DEFAULT_REGION=eu-central-1
AWS_ACCESS_KEY_ID=AKIAIOSFODNN7EXAMPLE # either via env or via ~/.aws
AWS_SECRET_ACCESS_KEY=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY # either via env or via ~/.aws
ENDPOINT=http://localhost:4566 # for the custom S3 storage endpoing (e.g. Minio or Localstack)

Or you can pass them via command line (see below)

Commands to run NEAR Lake, after ./target/release/near-lake

$ ./target/release/near-lake --home ~/.near/testnet run --stream-while-syncing --concurrency 50 sync-from-latest

After the network is synced, you should see logs of every block height currently received by NEAR Lake.

Whenever you run NEAR Lake for any network except localnet you'll need to sync with the network. This is required because it's a natural behavior of nearcore node and NEAR Lake is a wrapper for the regular nearcore node. In order to work and index the data your node must be synced with the network. This process can take a while, so we suggest to download a fresh backup of the data folder and put it in you --home-dir of your choice (by default it is ~/.near)

Running your NEAR Lake node on top of a backup data will reduce the time of syncing process because your node will download only the data after the backup was cut and it takes reasonable amount time.

All the backups can be downloaded from the public S3 bucket which contains latest daily snapshots:

You will need AWS CLI to be installed in order to download the backups.

$ aws s3 --no-sign-request cp s3://near-protocol-public/backups/mainnet/rpc/latest .
$ LATEST=$(cat latest)
$ aws s3 --no-sign-request cp --no-sign-request --recursive s3://near-protocol-public/backups/mainnet/rpc/$LATEST ~/.near/data

$ aws s3 --no-sign-request cp s3://near-protocol-public/backups/testnet/rpc/latest .
$ LATEST=$(cat latest)
$ aws s3 --no-sign-request cp --no-sign-request --recursive s3://near-protocol-public/backups/testnet/rpc/$LATEST ~/.near/data

It's not necessary but in order to index everything in the network it is better to do it from the genesis. nearcore node is running in non-archival mode by default. That means that the node keeps data only for 5 last epochs. In order to index data from the genesis we need to turn the node in archival mode.

To do it we need to update config.json located in --home-dir (by default it is ~/.near).

Find next keys in the config and update them as following:

{
  ...
  "archive": true,
  "tracked_shards": [0],
  ...
}

The syncing process in archival mode can take a lot of time, so it's better to download a backup provided by NEAR and put it in your data folder. After that your node will download only the data after the backup was cut and it takes reasonable amount time.

All the backups can be downloaded from the public S3 bucket which contains the latest daily snapshots:

• Archival Mainnet data folder
• Archival Testnet data folder

See this link for reference

We write all the data to AWS S3 buckets:

• near-lake-data-testnet (eu-central-1 region) for testnet
• near-lake-data-mainnet (eu-central-1 region) for mainnet

In case you want to run you own near-lake instance and store data in some S3 compatible storage (Minio or Localstack as example) You can owerride default S3 API endpoint by using --endpoint option

• run minio

$ mkdir -p /data/near-lake-custom && minio server /data

• run near-lake

$ ./target/release/near-lake --home ~/.near/testnet run --endpoint http://127.0.0.1:9000 --bucket near-lake-custom sync-from-latest

The data structure we use is the following:

<block_height>/
  block.json
  shard_0.json
  shard_1.json
  ...
  shard_N.json

• <block_height> is a 12-character-long u64 string with leading zeros (e.g 000042839521). See this issue for a reasoning
• block_json contains JSON-serialized BlockView struct. NB! this struct might change in the future, we will announce it
• shard_N.json where N is u64 starting from 0. Represents the index number of the shard. In order to find out the expected number of shards in the block you can look in block.json at .header.chunks_included

All NEAR Lake AWS S3 buckets have Request Payer enabled. It means that anyone with their own AWS credentials can List and Read the bucket's content and be charged for it by AWS. Connections to the bucket have to be done with AWS credentials provided. See NEAR Lake Framework for a reference.

Once we set up the public access to the buckets anyone will be able to build their own code to read it through.

For our own needs we are working on NEAR Lake Framework to have a simple way to create an indexer on top of the data stored by NEAR Lake itself.

See the official announce of NEAR Lake Framework on the NEAR Gov Forum

Here is a script that asks the RPC node about its peers and makes a list of boot nodes for you:

--boot-nodes `curl -X POST https://rpc.mainnet.near.org \
  -H "Content-Type: application/json" \
  -d '{
        "jsonrpc": "2.0",
        "method": "network_info",
        "params": [],
        "id": "dontcare"
      }' | \
jq '.result.active_peers as $list1 | .result.known_producers as $list2 |
$list1[] as $active_peer | $list2[] |
select(.peer_id == $active_peer.id) |
"\(.peer_id)@\($active_peer.addr)"' |\
awk 'NR>2 {print ","} length($0) {print p} {p=$0}' ORS="" | sed 's/"//g'