[go: up one dir, main page]

Skip to content

casper-network/casper-node

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

LOGO

casper-node

Reference node for the Casper Blockchain Protocol.

Casper Blockchain

Casper is the blockchain platform purpose-built to scale opportunity for everyone. Building toward blockchain’s next frontier, Casper is designed for real-world applications without sacrificing usability, cost, decentralization, or security. It removes the barriers that prevent mainstream blockchain adoption by making blockchain friendly to use, open to the world, and future-proof to support innovations today and tomorrow. Guided by open-source principles and built from the ground up to empower individuals, the team seeks to provide an equitable foundation made for long-lasting impact. Read more about our mission at: https://casper.network/network/casper-association

Current Development Status

The status on development is reported during the Community calls and is found here

The Casper MainNet is live.

Specification

Get Started with Smart Contracts

Community

Running a casper-node from source

Pre-Requisites for Building

  • CMake 3.1.4 or greater
  • Rust
  • libssl-dev
  • pkg-config
  • gcc
  • g++
  • recommended wasm-strip (used to reduce the size of compiled Wasm)
# Ubuntu prerequisites setup example
apt update
apt install cmake libssl-dev pkg-config gcc g++ -y
# the '-s -- -y' part ensures silent mode. Omit if you want to customize
curl https://sh.rustup.rs -sSf | sh -s -- -y

Setup

Before building a node, prepare your Rust build environment:

make setup-rs

The node software can be compiled afterwards:

cargo build -p casper-node --release

The result will be a casper-node binary found in target/release. Copy this somewhere into your PATH, or substitute target/release/casper-node for casper-node in all examples below.

Running one node

To run a validator node you will need to specify a config file and launch the validator subcommand, for example

casper-node validator /etc/casper-node/config.toml

The node ships with an example configuration file that should be setup first. There is also a template for a local chainspec in the same folder.

For launching, the following configuration values must be properly set:

Setting Description
network.known_addresses Must refer to public listening addresses of one or more currently-running nodes. If the node cannot connect to any of these addresses, it will panic. The node can be run with this referring to its own address, but it will be equivalent to specifying an empty list for known_addresses - i.e. the node will run and listen, but will be reliant on other nodes connecting to it in order to join the network. This would be normal for the very first node of a network, but all subsequent nodes should normally specify that first node's public listening address as their known_addresses.

The node will not run properly without another node to connect to. It is recommended that multiple nodes are run.

Running multiple nodes on one machine

There is a tool which automates the process of running multiple nodes on a single machine.

Note that running multiple nodes on a single machine is normally only recommended for test purposes.

Configuration

In general nodes are configured through a configuration file, typically named config.toml. This file may reference other files or locations through relative paths. When it does, note that all paths that are not absolute will be resolved relative to config.toml directory.

Environment overrides

Some environments may call for overriding options through the environment. In this scenario, the NODE_CONFIG environment variable can be used. For example: alternatively expressed as

export NODE_CONFIG=consensus.secret_key_path=secret_keys/node-1.pem;network.known_addresses=[1.2.3.4:34553, 200.201.203.204:34553]
casper-node validator /etc/casper-node/config.toml

Note how the semicolon is used to separate configuration overrides here.

Other environment variables

To set the threshold at which a warn-level log message is generated for a long-running reactor event, use the env var CL_EVENT_MAX_MICROSECS. For example, to set the threshold to 1 millisecond:

CL_EVENT_MAX_MICROSECS=1000

To set the threshold above which the size of the current scheduler queues will be dumped to logs, use the CL_EVENT_QUEUE_DUMP_THRESHOLD variable. For example, to set the threshold to 10000 events:

CL_EVENT_QUEUE_DUMP_THRESHOLD=10000

This will dump a line to the log if the total number of events in queues exceeds 10000. After each dump, the threshold will be automatically increased by 10% to avoid log flooding.

Example log entry:

Current event queue size (11000) is above the threshold (10000): details [("FinalitySignature", 3000), ("FromStorage", 1000), ("NetworkIncoming", 6500), ("Regular", 500)]

Logging

Logging can be enabled by setting the environment variable RUST_LOG. This can be set to one of the following levels, from lowest priority to highest: trace, debug, info, warn, error:

RUST_LOG=info cargo run --release -- validator resources/local/config.toml

If the environment variable is unset, it is equivalent to setting RUST_LOG=error.

Log message format

A typical log message will look like:

Jun 09 01:40:17.315 INFO  [casper_node::components::rpc_server rpc_server.rs:127] starting HTTP server; server_addr=127.0.0.1:7777

This is comprised of the following parts:

  • timestamp
  • log level
  • full module path (not to be confused with filesystem path) of the source of the message
  • filename and line number of the source of the message
  • message

Filtering log messages

RUST_LOG can be set to enable varying levels for different modules. Simply set it to a comma-separated list of module-path=level, where the module path is as shown above in the typical log message, with the end truncated to suit.

For example, to enable trace level logging for the network module in components, info level for all other modules in components, and warn level for the remaining codebase:

RUST_LOG=casper_node::components::network=trace,casper_node::comp=info,warn

Logging network messages and tracing events

Special logging targets exist in net_in and net_out which can be used to log every single network message leaving or entering a node when set to trace level:

RUST_LOG=net_in::TRACE,net_out::TRACE

All messages in these logs are also assigned a unique ID that is different even if the same message is sent to multiple nodes. The receiving node will log them using the same ID as the sender, thus enabling the tracing of a message across multiple nodes provided all logs are available.

Another helpful logging feature is ancestor logging. If the target dispatch is set to at least debug level, events being dispatched will be logged as well. Any event has an id (ev) and may have an ancestor (a), which is the previous event whose effects caused the resulting event to be scheduled. As an example, if an incoming network message gets assigned an ID of ev=123, the first round of subsequent events will show a=123 as their ancestor in the logs.

Changing the logging filter at runtime

If necessary, the filter of a running node can be changed using the diagnostics port, using the set-log-filter command. See the "Diagnostics port" section for details on how to access it.

Debugging

Some additional debug functionality is available, mainly allowed for inspections of the internal event queue.

Diagnostics port

If the configuration option diagnostics_port.enabled is set to true, a unix socket named debug.socket by default can be found next to the configuration while the node is running.

Interactive use

The debug.socket can be connected to by tools like socat for interactive use:

socat readline unix:/path/to/debug.socket

Entering help will show available commands. The set command allows configuring the current connection, see set --help.

Example: Collecting a consensus dump

After connecting using socat (see above), we set the output format to JSON:

set --output=json

A confirmation will acknowledge the settings change (unless --quiet=true is set):

{
  "Success": {
    "msg": "session unchanged"
  }
}

We can now call dump-consensus to get the latest era serialized in JSON format:

dump-consensus
{
  "Success": {
    "msg": "dumping consensus state"
  }
}
{"id":8,"start_time":"2022-03-01T14:54:42.176Z","start_height":88,"new_faulty" ...

An era other than the latest can be dumped by specifying as a parameter, e.g. dump-consensus 3 will dump the third era. See dump-consensus --help for details.

Example: Dumping the event queue

With the connection set to JSON output (see previous example), we can also dump the event queues:

dump-queues
{
  "Success": {
    "msg": "dumping queues"
  }
}
{"queues":{"Regular":[],"Api":[],"Network":[],"Control":[],"NetworkIncoming":[]
}}{"queues":{"Api":[],"Regular":[],"Control":[],"NetworkIncoming":[],"Network":
[]}}{"queues":{"Network":[],"Control":[],"Api":[],"NetworkIncoming":[],"Regular
":[]}}

Empty output will be produced on a node that is working without external pressure, as the queues will be empty most of the time.

Non-interactive use

The diagnostics port can also be scripted by sending a newline-terminated list of commands through socat. For example, the following sequence of commands will collect a consensus dump without the success-indicating header:

set -o json -q true
dump-consensus

For ad-hoc dumps, this can be shortened and piped into socat:

echo -e 'set -o json -q true\ndump-consensus' | socat - unix-client:debug.socket > consensus-dump.json

This results in the latest era being dumped into consensus-dump.json.

Running a client

See the client README.

Running a local network

See the nctl utility README.

Running on an existing network

To support upgrades with a network, the casper-node is installed using scripts distributed with the casper-node-launcher.