Skip to content

A Kurtosis package that deploys a private, portable, and modular Ethereum devnet

License

Notifications You must be signed in to change notification settings

Tofel/ethereum-package

 
 

Repository files navigation

Ethereum Package

Run of the Ethereum Network Package

This is a Kurtosis package that will spin up a private Ethereum testnet over Docker or Kubernetes with multi-client support, Flashbot's mev-boost infrastructure for PBS-related testing/validation, and other useful network tools (transaction spammer, monitoring tools, etc). Kurtosis packages are entirely reproducible and composable, so this will work the same way over Docker or Kubernetes, in the cloud or locally on your machine.

Specifically, this package will:

  1. Generate Execution Layer (EL) & Consensus Layer (CL) genesis information using the Ethereum genesis generator.
  2. Configure & bootstrap a network of Ethereum nodes of n size using the genesis data generated above
  3. Spin up a transaction spammer to send fake transactions to the network
  4. Spin up and connect a testnet verifier
  5. Spin up a Grafana and Prometheus instance to observe the network

Optional features (enabled via flags or parameter files at runtime):

  • Block until the Beacon nodes finalize an epoch (i.e. finalized_epoch > 0)
  • Spin up & configure parameters for the infrastructure behind Flashbot's implementation of PBS using mev-boost, in either full or mock mode. More details here.
  • Spin up & connect the network to a beacon metrics gazer service to collect network-wide participation metrics.
  • Spin up and connect a JSON RPC Snooper to the network log responses & requests between the EL engine API and the CL client.
  • Specify extra parameters to be passed in for any of the: CL client Beacon, and CL client validator, and/or EL client containers
  • Specify the required parameters for the nodes to reach an external block building network
  • Generate keystores for each node in parallel

Quickstart

Open in Gitpod

  1. Install Docker & start the Docker Daemon if you haven't done so already

  2. Install the Kurtosis CLI, or upgrade it to the latest version if it's already installed

  3. Run the package with default configurations from the command line:

    kurtosis run --enclave my-testnet github.com/kurtosis-tech/ethereum-package

Run with your own configuration

Kurtosis packages are parameterizable, meaning you can customize your network and its behavior to suit your needs by storing parameters in a file that you can pass in at runtime like so:

kurtosis run --enclave my-testnet github.com/kurtosis-tech/ethereum-package "$(cat ~/network_params.yaml)"

Where network_params.yaml contains the parameters for your network in your home directory.

Run on Kubernetes

Kurtosis packages work the same way over Docker or on Kubernetes. Please visit our Kubernetes docs to learn how to spin up a private testnet on a Kubernetes cluster.

Tear down

The testnet will reside in an enclave - an isolated, ephemeral environment. The enclave and its contents (e.g. running containers, files artifacts, etc) will persist until torn down. You can remove an enclave and its contents with:

kurtosis enclave rm -f my-testnet

Management

The Kurtosis CLI can be used to inspect and interact with the network.

For example, if you need shell access, simply run:

kurtosis service shell my-testnet $SERVICE_NAME

And if you need the logs for a service, simply run:

kurtosis service logs my-testnet $SERVICE_NAME

Check out the full list of CLI commands here

Debugging

To grab the genesis files for the network, simply run:

kurtosis files download my-testnet $FILE_NAME $OUTPUT_DIRECTORY

For example, to retrieve the Execution Layer (EL) genesis data, run:

kurtosis files download my-testnet el-genesis-data ~/Downloads

Configuration

To configure the package behaviour, you can modify your network_params.yaml file. The full YAML schema that can be passed in is as follows with the defaults provided:

Click to show all configuration options
# Specification of the participants in the network
participants:
  # The type of EL client that should be started
  # Valid values are geth, nethermind, erigon, besu, ethereumjs, reth
- el_client_type: geth

  # The Docker image that should be used for the EL client; leave blank to use the default for the client type
  # Defaults by client:
  # - geth: ethereum/client-go:latest
  # - erigon: thorax/erigon:devel
  # - nethermind: nethermind/nethermind:latest
  # - besu: hyperledger/besu:develop
  # - reth: ghcr.io/paradigmxyz/reth
  # - ethereumjs: ethpandaops/ethereumjs:master
  el_client_image: ""

  # The log level string that this participant's EL client should log at
  # If this is emptystring then the global `logLevel` parameter's value will be translated into a string appropriate for the client (e.g. if
  # global `logLevel` = `info` then Geth would receive `3`, Besu would receive `INFO`, etc.)
  # If this is not emptystring, then this value will override the global `logLevel` setting to allow for fine-grained control
  # over a specific participant's logging
  el_client_log_level: ""

  # A list of optional extra params that will be passed to the EL client container for modifying its behaviour
  el_extra_params: []

  # A list of optional extra env_vars the el container should spin up with
  el_extra_env_vars: {}

  # The type of CL client that should be started
  # Valid values are nimbus, lighthouse, lodestar, teku, and prysm
  cl_client_type: lighthouse

  # The Docker image that should be used for the EL client; leave blank to use the default for the client type
  # Defaults by client (note that Prysm is different in that it requires two images - a Beacon and a validator - separated by a comma):
  # - lighthouse: sigp/lighthouse:latest
  # - teku: consensys/teku:latest
  # - nimbus: statusim/nimbus-eth2:multiarch-latest
  # - prysm: gcr.io/prysmaticlabs/prysm/beacon-chain:latest,gcr.io/prysmaticlabs/prysm/validator:latest
  # - lodestar: chainsafe/lodestar:next
  cl_client_image: ""

  # The log level string that this participant's EL client should log at
  # If this is emptystring then the global `logLevel` parameter's value will be translated into a string appropriate for the client (e.g. if
  # global `logLevel` = `info` then Teku would receive `INFO`, Prysm would receive `info`, etc.)
  # If this is not emptystring, then this value will override the global `logLevel` setting to allow for fine-grained control
  # over a specific participant's logging
  cl_client_log_level: ""

  # A list of optional extra params that will be passed to the CL client Beacon container for modifying its behaviour
  # If the client combines the Beacon & validator nodes (e.g. Teku, Nimbus), then this list will be passed to the combined Beacon-validator node
  beacon_extra_params: []

  # A list of optional extra params that will be passed to the CL client validator container for modifying its behaviour
  # If the client combines the Beacon & validator nodes (e.g. Teku, Nimbus), then this list will also be passed to the combined Beacon-validator node
  validator_extra_params: []

  # A set of parameters the node needs to reach an external block building network
  # If `null` then the builder infrastructure will not be instantiated
  # Example:
  #
  # "relay_endpoints": [
  #  "https:#[email protected]",
  #  "https:#[email protected]",
  #  "https:#[email protected]",
  #  "https:#[email protected]"
  # ]
  builder_network_params: null

  # Resource management for el/beacon/validator containers
  # CPU is milicores
  # RAM is in MB
  # Defaults are set per client
  el_min_cpu: 0
  el_max_cpu: 0
  el_min_mem: 0
  el_max_mem: 0
  bn_min_cpu: 0
  bn_max_cpu: 0
  bn_min_mem: 0
  bn_max_mem: 0
  v_min_cpu: 0
  v_max_cpu: 0
  v_min_mem: 0
  v_max_mem: 0

  # Snooper can be enabled with the `snooper_enabled` flag per client or globally
  # Defaults to false
  snooper_enabled: false

  # Enables Ethereum Metrics Exporter for this participant. Can be set globally.
  # Defaults to false
  ethereum_metrics_exporter_enabled: false

  # Count of nodes to spin up for this participant
  # Default to 1
  count: 1

  # Count of the number of validators you want to run for a given participant
  # Default to null, which means that the number of validators will be using the
  # network parameter num_validator_keys_per_node
  validator_count: null

  # Prometheus additional configuration for a given participant prometheus target.
  # Execution, beacon and validator client targets on prometheus will include this
  # configuration.
  prometheus_config:
    # Scrape interval to be used. Default to 15 seconds
    scrape_interval: 15s
    # Additional labels to be added. Default to empty
    labels: {}

# Default configuration parameters for the Eth network
network_params:
  # The network ID of the network.
  network_id: 3151908

  # The address of the staking contract address on the Eth1 chain
  deposit_contract_address: "0x4242424242424242424242424242424242424242"

  # Number of seconds per slot on the Beacon chain
  seconds_per_slot: 12

  # The number of validator keys that each CL validator node should get
  num_validator_keys_per_node: 64

  # This mnemonic will a) be used to create keystores for all the types of validators that we have and b) be used to generate a CL genesis.ssz that has the children
  # validator keys already preregistered as validators
  preregistered_validator_keys_mnemonic: "giant issue aisle success illegal bike spike question tent bar rely arctic volcano long crawl hungry vocal artwork sniff fantasy very lucky have athlete"
  # How long you want the network to wait before starting up
  genesis_delay: 120

  # Max churn rate for the network introduced by
  # EIP-7514 https:#eips.ethereum.org/EIPS/eip-7514
  # Defaults to 8
  max_churn: 8

  # Ejection balance
  # Defaults to 16ETH
  # 16000000000 gwei
  ejection_balance: 16000000000,

  # The epoch at which the capella and deneb forks are set to occur.
  capella_fork_epoch: 0
  deneb_fork_epoch: 500
  electra_fork_epoch: null

# Configuration place for transaction spammer - https:#github.com/MariusVanDerWijden/tx-fuzz
tx_spammer_params:
  # A list of optional extra params that will be passed to the TX Spammer container for modifying its behaviour
  tx_spammer_extra_args: []

# Configuration place for goomy the blob spammer - https:#github.com/ethpandaops/goomy-blob
goomy_blob_params:
  # A list of optional params that will be passed to the blob-spammer comamnd for modifying its behaviour
  goomy_blob_args: []

# By default includes
# - A transaction spammer & blob spammer is launched to fake transactions sent to the network
# - Forkmon for EL will be launched
# - A prometheus will be started, coupled with grafana
# - A beacon metrics gazer will be launched
# - A light beacon chain explorer will be launched
# - Default: ["tx_spammer", "blob_spammer", "el_forkmon", "beacon_metrics_gazer", "dora"," "prometheus_grafana"]
additional_services:
  - broadcaster
  - tx_spammer
  - blob_spammer
  - custom_flood
  - goomy_blob
  - el_forkmon
  - beacon_metrics_gazer
  - dora
  - full_beaconchain_explorer
  - prometheus_grafana

# If set, the package will block until a finalized epoch has occurred.
wait_for_finalization: false

# The global log level that all clients should log at
# Valid values are "error", "warn", "info", "debug", and "trace"
# This value will be overridden by participant-specific values
global_client_log_level: "info"

# EngineAPI Snooper global flags for all participants
# Default to false
snooper_enabled: false

# Enables Ethereum Metrics Exporter for all participants
# Defaults to false
ethereum_metrics_exporter_enabled: false

# Parallelizes keystore generation so that each node has keystores being generated in their own container
# This will result in a large number of containers being spun up than normal. We advise users to only enable this on a sufficiently large machine or in the cloud as it can be resource consuming on a single machine.
parallel_keystore_generation: false

# Disable peer scoring to prevent nodes impacted by faults from being permanently ejected from the network
# Default to false
disable_peer_scoring: false

# A list of locators for grafana dashboards to be loaded be the grafana service
grafana_additional_dashboards: []

# Supports three valeus
# Default: "null" - no mev boost, mev builder, mev flood or relays are spun up
# "mock" - mock-builder & mev-boost are spun up
# "full" - mev-boost, relays, flooder and builder are all spun up
# Users are recommended to set network_params.capella_fork_epoch to non zero when testing MEV
# We have seen instances of multibuilder instances failing to start mev-relay-api with non zero epochs
mev_type: null

# Parameters if MEV is used
mev_params:
  # The image to use for MEV boot relay
  mev_relay_image: flashbots/mev-boost-relay
  # The image to use for the builder
  mev_builder_image: ethpandaops/flashbots-builder:main
  # The image to use for the CL builder
  mev_builder_cl_image: sigp/lighthouse:latest
  # The image to use for mev-boost
  mev_boost_image: flashbots/mev-boost
  # Extra parameters to send to the API
  mev_relay_api_extra_args: []
  # Extra parameters to send to the housekeeper
  mev_relay_housekeeper_extra_args: []
  # Extra parameters to send to the website
  mev_relay_website_extra_args: []
  # Extra parameters to send to the builder
  mev_builder_extra_args: []
  # Prometheus additional configuration for the mev builder participant.
  # Execution, beacon and validator client targets on prometheus will include this configuration.
  mev_builder_prometheus_config:
    # Scrape interval to be used. Default to 15 seconds
    scrape_interval: 15s
    # Additional labels to be added. Default to empty
    labels: {}
  # Image to use for mev-flood
  mev_flood_image: flashbots/mev-flood
  # Extra parameters to send to mev-flood
  mev_flood_extra_args: []
  # Number of seconds between bundles for mev-flood
  mev_flood_seconds_per_bundle: 15
  # Optional parameters to send to the custom_flood script that sends reliable payloads
  custom_flood_params:
    interval_between_transactions: 1

Example configurations

Verkle configuration example
participants:
  - el_client_type: geth
    el_client_image: ethpandaops/geth:<VERKLE_IMAGE>
    elExtraParams:
    - "--override.verkle=<UNIXTIMESTAMP>"
    cl_client_type: lighthouse
    cl_client_image: sigp/lighthouse:latest
  - el_client_type: geth
    el_client_image: ethpandaops/geth:<VERKLE_IMAGE>
    elExtraParams:
    - "--override.verkle=<UNIXTIMESTAMP>"
    cl_client_type: lighthouse
    cl_client_image: sigp/lighthouse:latest
  - el_client_type: geth
    el_client_image: ethpandaops/geth:<VERKLE_IMAGE>
    elExtraParams:
    - "--override.verkle=<UNIXTIMESTAMP>"
    cl_client_type: lighthouse
    cl_client_image: sigp/lighthouse:latest
network_params:
  capella_fork_epoch: 2
  deneb_fork_epoch: 5
additional_services: []
wait_for_finalization: false
wait_for_verifications: false
global_client_log_level: info
A 3-node Ethereum network with "mock" MEV mode. Useful for testing mev-boost and the client implementations without adding the complexity of the relay. This can be enabled by a single config command and would deploy the [mock-builder](https://github.com/marioevz/mock-builder), instead of the relay infrastructure.
participants:
  - el_client_type: geth
    el_client_image: ''
    cl_client_type: lighthouse
    cl_client_image: ''
    count: 2
  - el_client_type: nethermind
    el_client_image: ''
    cl_client_type: teku
    cl_client_image: ''
    count: 1
  - el_client_type: besu
    el_client_image: ''
    cl_client_type: prysm
    cl_client_image: ''
    count: 2
mev_type: mock
additional_services: []
A 5-node Ethereum network with three different CL and EL client combinations and mev-boost infrastructure in "full" mode.
participants:
  - el_client_type: geth
    cl_client_type: lighthouse
    count: 2
  - el_client_type: nethermind
    cl_client_type: teku
  - el_client_type: besu
    cl_client_type: prysm
    count: 2
mev_type: full
network_params:
  capella_fork_epoch: 1
additional_services: []
A 2-node geth/lighthouse network with optional services (Grafana, Prometheus, transaction-spammer, EngineAPI snooper, and a testnet verifier)
participants:
  - el_client_type: geth
    cl_client_type: lighthouse
    count: 2
snooper_enabled: true

Custom labels for Docker and Kubernetes

There are 4 custom labels that can be used to identify the nodes in the network. These labels are used to identify the nodes in the network and can be used to run chaos tests on specific nodes. An example for these labels are as follows:

Execution Layer (EL) nodes:

  "com.kurtosistech.custom.ethereum-package-client": "geth",
  "com.kurtosistech.custom.ethereum-package-client-image": "ethereum-client-go-latest",
  "com.kurtosistech.custom.ethereum-package-client-type": "execution",
  "com.kurtosistech.custom.ethereum-package-connected-client": "lighthouse",

Consensus Layer (CL) nodes - Beacon:

  "com.kurtosistech.custom.ethereum-package-client": "lighthouse",
  "com.kurtosistech.custom.ethereum-package-client-image": "sigp-lighthouse-latest",
  "com.kurtosistech.custom.ethereum-package-client-type": "beacon",
  "com.kurtosistech.custom.ethereum-package-connected-client": "geth",

Consensus Layer (CL) nodes - Validator:

  "com.kurtosistech.custom.ethereum-package-client": "lighthouse",
  "com.kurtosistech.custom.ethereum-package-client-image": "sigp-lighthouse-latest",
  "com.kurtosistech.custom.ethereum-package-client-type": "validator",
  "com.kurtosistech.custom.ethereum-package-connected-client": "geth",

ethereum-package-client describes which client is running on the node. ethereum-package-client-image describes the image that is used for the client. ethereum-package-client-type describes the type of client that is running on the node (execution,beacon or validator). ethereum-package-connected-client describes the CL/EL client that is connected to the EL/CL client.

Proposer Builder Separation (PBS) emulation

To spin up the network of Ethereum nodes with an external block building network (using Flashbot's mev-boost protocol), simply use:

kurtosis run github.com/kurtosis-tech/ethereum-package '{"mev_type": "full"}'

Starting your network up with "mev_type": "full" will instantiate and connect the following infrastructure to your network:

  1. Flashbot's block builder & CL validator + beacon - A modified Geth client that builds blocks. The CL validator and beacon clients are lighthouse clients configured to receive payloads from the relay.
  2. mev-relay-api - Services that provide APIs for (a) proposers, (b) block builders, (c) data
  3. mev-relay-website - A website to monitor payloads that have been delivered
  4. mev-relay-housekeeper - Updates known validators, proposer duties, and more in the background. Only a single instance of this should run.
  5. mev-boost - open-source middleware instantiated for each EL/Cl pair in the network, including the builder
  6. mev-flood - Deploys UniV2 smart contracts, provisions liquidity on UniV2 pairs, & sends a constant stream of UniV2 swap transactions to the network's public mempool.
Caveats when using "mev_type": "full"
  • Validators (64 per node by default, so 128 in the example in this guide) will get registered with the relay automatically after the 1st epoch. This registration process is simply a configuration addition to the mev-boost config - which Kurtosis will automatically take care of as part of the set up. This means that the mev-relay infrastructure only becomes aware of the existence of the validators after the 1st epoch.
  • After the 3rd epoch, the mev-relay service will begin to receive execution payloads (eth_sendPayload, which does not contain transaction content) from the mev-builder service (or mock-builder in mock-mev mode).
  • Validators will start to receive validated execution payload headers from the mev-relay service (via mev-boost) after the 4th epoch. The validator selects the most valuable header, signs the payload, and returns the signed header to the relay - effectively proposing the payload of transactions to be included in the soon-to-be-proposed block. Once the relay verifies the block proposer's signature, the relay will respond with the full execution payload body (incl. the transaction contents) for the validator to use when proposing a SignedBeaconBlock to the network.

It is recommended to use non zero value for capella_fork_epoch by setting network_params.capella_fork_epoch to a non-zero value in the arguments passed with mev_type set to full.

This package also supports a "mev_type": "mock" mode that will only bring up:

  1. mock-builder - a server that listens for builder API directives and responds with payloads built using an execution client
  2. mev-boost - for every EL/CL pair launched

For more details, including a guide and architecture of the mev-boost infrastructure, go here.

MEV-Boost usage with Capella at Epoch 0

This note is from 2023-10-05

flashbots/mev-boost-relay:0.27 and later support capella_fork_epoch at 0 but this seems to require a few flags enabled on the lighthouse beacon client including --always-prefer-builder-payload and --disable-peer-scoring

Users are recommended to use examples/capella-mev.yaml; as inspiration for reliable payload delivery.

Pre-funded accounts at Genesis

This package comes with seven prefunded keys for testing.

Here's a table of where the keys are used

Account Index Component Used In Private Key Used Public Key Used Comment
0 Builder As coinbase
0 mev_custom_flood As the receiver of balance
1 blob_spammer As the sender of blobs
3 transaction_spammer To spam transactions with
4 goomy_blob As the sender of blobs
5 eip4788_deployment As contract deployer
6 mev_flood As the contract owner
7 mev_flood As the user_key
11 mev_custom_flood As the sender of balance

Developing On This Package

First, install prerequisites:

  1. Install Kurtosis itself

Then, run the dev loop:

  1. Make your code changes

  2. Rebuild and re-run the package by running the following from the root of the repo:

    kurtosis run . "{}"

    NOTE 1: You can change the value of the second positional argument flag to pass in extra configuration to the package per the "Configuration" section above! NOTE 2: The second positional argument accepts JSON.

To get detailed information about the structure of the package, visit the architecture docs.

When you're happy with your changes:

  1. Create a PR
  2. Add one of the maintainers of the repo as a "Review Request":
    • parithosh (Ethereum Foundation)
    • barnabasbusa (Ethereum Foundation)
    • pk910 (Ethereum Foundation)
    • samcm (Ethereum Foundation)
    • h4ck3rk3y (Kurtosis)
    • mieubrisse (Kurtosis)
    • leederek (Kurtosis)
  3. Once everything works, merge!

About

A Kurtosis package that deploys a private, portable, and modular Ethereum devnet

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Starlark 97.1%
  • Python 2.9%