Skip to content

Latest commit

 

History

History
347 lines (248 loc) · 13.1 KB

File metadata and controls

347 lines (248 loc) · 13.1 KB

IMPORTANT: See Prerequisites below

Digital Asset RPC API Infrastructure

This repo houses the API Ingester and Database Types components of the Metaplex Digital Asset RPC API. Together these components are responsible for the aggregation of Solana Validator Data into an extremely fast and well typed api. This api provides a nice interface on top of the Metaplex programs. It abstracts the byte layout on chain, allows for super-fast querying and searching, as well as serves the merkle proofs needed to operate over compressed nfts.

Components

  1. Ingester -> A background processing system that gets messages from a Messenger, and uses BlockBuster Parsers to store the canonical representation of Metaplex types in a storage system. This system also holds the re-articulated Merkle tree that supports the compressed NFTs system.
  2. Api -> A JSON Rpc api that serves Metaplex objects. This api allows filtering, pagination and searching over Metaplex data. This data includes serving the merkle proofs for the compressed NFTs system. It is intended to be run right alongside the Solana RPC and works in much the same way. Just like the solana RPC takes data from the validator and serves it in a new format, so this api takes data off the validator and serves it.

The API specification is located here https://github.com/metaplex-foundation/api-specifications This spec is what providers of this api must implement against.

Infrastructure and Deployment Examples

Along with the above rust binaries, this repo also maintains examples and best practice settings for running the entire infrastructure. The example infrastructure is as follows.

  • A Solana No-Vote Validator - This validator is configured to only have secure access to the validator ledger and account data under consensus.
  • A Geyser Plugin (Plerkle) - The above validator is further configured to load this geyser plugin that sends Plerkle Serialized Messages over a messaging system.
  • A Redis Cluster (Stream Optimized) - The example messaging system is a light weight redis deployment that supports the streaming configuration.
  • A Kubernetes Cluster - The orchestration system for the API and Ingester processes. Probably overkill for a small installation, but it's a rock solid platform for critical software.

This repo houses Helm Charts, Docker files and Terraform files to assist in the deployment of the example infrastructure.

Developing

Prerequisites:

You must clone the https://github.com/metaplex-foundation/blockbuster repo, this is un publishable for now due to active development in like 1000 branches and serious mathematics avoiding dependency hell.

Because this is a multi component system the easiest way to develop or locally test this system is with docker but developing locally without docker is possible.

Regenerating DB Types

Edit the init.sql, then run docker compose up db Then with a local DATABASE_URL var exported like this export DATABASE_URL=postgres://solana:solana@localhost/solana you can run sea-orm-cli generate entity -o ./digital_asset_types/src/dao/generated/ --database-url $DATABASE_URL --with-serde both --expanded-format

If you need to install sea-orm-cli run cargo install sea-orm-cli.

Note: The current SeaORM types were generated using version 0.9.3 so unless you want to upgrade you can install using cargo install sea-orm-cli --version 0.9.3.

Developing Locally

Prerequisites

  • A Postgres Server running with the database setup according to ./init.sql
  • A Redis instance that has streams enabled or a version that supports streams
  • A local solana validator with the Plerkle plugin running.
  • Environment Variables set to allow your validator, ingester and api to access those prerequisites.

See Plugin Configuration for how to locally configure the test validator plugin to work.

For the API you need the following environment variables:

APP_DATABASE_URL=postgres://solana:solana@db/solana  #change to your db host
APP_SERVER_PORT=9090
cargo run -p das_api

For the Ingester you need the following environment variables:

INGESTER_DATABASE_CONFIG: '{listener_channel="backfill_item_added", url="postgres://solana:solana@db/solana"}' # your database host
INGESTER_MESSENGER_CONFIG: '{messenger_type="Redis", connection_config={ redis_connection_str="redis://redis" } }' #your redis
INGESTER_RPC_CONFIG: '{url="http://validator:8899", commitment="finalized"}' # your solana validator or same network rpc, if local you must use your solana instance running localy
cargo run -p nft_ingester

When making changes you will need to stop the cargo process and re-run. Someday we will have auto rebuild for local cargo stuff but for now you are on your own.

NOTE

INGESTER_ROLE

This environment variable can be used to split the work load.

All for a combined setup Ingester for just the Listeners to txn and acct Backfiller for just the backfiller scheduler and notifyer Background for just the background tasks.

For production you should split the coponents up.

Developing With Docker

Developing with Docker is much easier, but has some nuances to it. This test docker compose system relies on a programs folder being accessible, this folder needs to have the shared object files for the following programs

  • Token Metadata
  • Bubblegum
  • Gummyroll
  • Token 2022
  • Latest version of the Associated token program

You need to run the following script in order to get the .so files.

./prepare-local-docker-env.sh

This script grabs all the code for these programs and compiles it, and chucks it into your programs folder. Go grab some coffe because this will take a while/ If you get some permissions errors, just sudo delete the programs directory and start again.

Authentication with Docker and AWS

aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin {your aws container registry}

Running the application

We use docker-compose to build the multi-container Docker application. On some systems its docker compose.

docker-compose build

This builds the docker container for API and the Ingester components and will download the appropriate Redis, Postgres and Solana+plerkle docker images. Keep in mind that the version latest on the Solana Validator image will match the latest version available on the docs, for other versions please change that version in your docker compose file.

docker-compose up

Developing

When making changes you will need to docker compose up --build --force-recreate again to get the latest changes. Also when mucking about with the docker file if your gut tells you that something is wrong, and you are getting build errors run docker compose build --no-cache

Sometimes you will want to delete the db do so with sudo rm -rf db-data. You can also delete the ledger with sudo rm -rf ledger.

Running Bubblegum Test Sequences

While running the multi-container Docker application locally, you can run a script located in tools/txn_forwarder/bubblegum_tests that will send sequences of bubblegum transactions via the txn_forwarder, and then use psql to read and verify the indexing results in the local Postgres database.

sudo rm -rf db-data/
sudo rm -rf ledger/
docker compose up --force-recreate --build

In another terminal:

cd tools/txn_forwarder/bubblegum_tests/
./run-bubblegum-sequences.sh

You should see it log something like:

Running 10 scenarios forwards
mint_transfer_burn.scenario initial asset table state passed
mint_transfer_burn.scenario initial asset_creators table state passed
mint_transfer_burn.scenario initial asset_grouping table state passed
mint_transfer_burn.scenario initial cl_items table state passed
...
mint_to_collection_unverify_collection.scenario asset table passed
mint_to_collection_unverify_collection.scenario asset_creators table passed
mint_to_collection_unverify_collection.scenario asset_grouping table passed
mint_to_collection_unverify_collection.scenario cl_items table passed

ALL TESTS PASSED FORWARDS!

You can also run the sequences in reverse:

./run-bubblegum-sequences.sh reverse

And after it runs you should see ALL TESTS PASSED IN REVERSE!

A few detailed notes about this test script:

  • This script is not all-encompassing. It is only meant to automate some normal basic tests that were previously done manually. The reason this test is not added to CI is because requires a more powerful system to run the Docker application, which contains the no-vote Solana validator.
  • The test sequences are in .scenario files, but instead of sending those files to the txn_forwarder directly (which supports the file format), we parse them out and send them individually using the single parameter. This is because using the .scenario file directly results in random ordering of the transactions and we are explicity trying to test them going forwards and in reverse.
  • In general the expected database results are the same when running the transactions forwards and backwards. However, for assets that are decompressed, this is not true because we don't index some of the asset information from Bubblegum mint indexing if we already know the asset has been decompressed. We instead let Token Metadata account based indexing fill in that information. This is not reflected by this test script so the results differ when running these sequences in reverse. The differing results are reflected in test files with the _reverse suffix.

Logs

To get a reasonable amount of logs while running Docker, direct grafana logs to a file:

grafana:
    ...
    environment:
      ...
      - GF_LOG_MODE=file

and set Solana Rust logs to error level:

  solana:
    ...
    environment:
      RUST_LOG: error

Interacting with API

Once everything is working you can see that there is a api being served on

http://localhost:9090

And a Metrics System on

http://localhost:3000

Here are some example requests to the Read API:

curl --request POST --url http://localhost:9090 --header 'Content-Type: application/json' --data '{
    "jsonrpc": "2.0",
    "method": "getAssetsByOwner",
    "params": [
      "CMvMqPNKHikuGi7mrngvQzFeQ4rndDnopx3kc9drne8M",
      { "sortBy": "created", "sortDirection": "asc"},
      50,
      1,
      "",
      ""
    ],
    "id": 0
}' | json_pp

curl --request POST --url http://localhost:9090 --header 'Content-Type: application/json' --data '{
    "jsonrpc": "2.0",
    "method": "getAsset",
    "params": [
      "8vw7tdLGE3FBjaetsJrZAarwsbc8UESsegiLyvWXxs5A"
    ],
    "id": 0
}' | json_p

curl --request POST --url http://localhost:9090 --header 'Content-Type: application/json' --data '{
    "jsonrpc": "2.0",
    "method": "getAssetProof",
    "params": [
      "8vw7tdLGE3FBjaetsJrZAarwsbc8UESsegiLyvWXxs5A"
    ],
    "id": 0
}' | json_pp

Deploying to Kubernetes

Using skaffold you can deploy to k8s, make sure you authenticate with your docker registry

Make sure you have the env vars you need to satisfy this part of the skaffold.yaml

...
    setValueTemplates:
      ingest.db_url: "{{.DATABASE_URL}}"
      ingest.rpc_url: "{{.RPC_URL}}"
      ingest.redis_url: "{{.REDIS_URL}}"
      metrics.data_dog_api_key: "{{.DATA_DOG_API}}"
      load.seed: "{{.LOAD_SEED}}"
      load.rpc_url: "{{.RPC_URL}}"
      valuesFiles:
        - ./helm/ingest/values.yaml
  - name: das-api
    chartPath: helm/api
    artifactOverrides:
      image: public.ecr.aws/k2z7t6t6/metaplex-rpc-api
    setValueTemplates:
      api.db_url: "{{.DATABASE_URL}}"
      api.redis_url: "{{.REDIS_URL}}"
      metrics.data_dog_api_key: "{{.DATA_DOG_API}}"
...
skaffold build --file-output skaffold-state.json --cache-artifacts=false
## Your namepsace may differ.
skaffold deploy -p devnet --build-artifacts skaffold-state.json --namespace devnet-read-api --tail=true

METRICS

Here are the metrics that various parts of ths system expose;

NFT INGESTER

ACKING

count ingester.ack - number of messages acked tagged by stream

count ingester.stream.ack_error - error acking a message count ingester.stream.receive_error - error getting stream data

Stream Metrics

ingester.stream_redelivery - Stream tagged of messages re delivered ingester.stream_size - Size of stream, tagged by stream ingester.stream_size_error - Error getting the stream size

Stream Specific Metrics

All these metrics are tagged by stream count ingester.seen time ingester.proc_time count ingester.ingest_success count ingester.ingest_redeliver_success count ingester.not_implemented count ingester.ingest_error

BG Tasks

time ingester.bgtask.proc_time count ingester.bgtask.success count ingester.bgtask.error count ingester.bgtask.network_error time ingester.bgtask.bus_time count ingester.bgtask.identical

BACKFILLER

count ingester.backfiller.task_panic count ingester.backfiller.task_error guage ingester.backfiller.missing_trees

Startup

ingester.startup

API

api_call