Configuration
CoW Protocol Services employ a hierarchical configuration system supporting command-line arguments, environment variables, TOML files, and runtime adjustments. All services include --help documentation.
Configuration Priority
Settings follow this precedence (highest to lowest):
- Command-line arguments
- Environment variables
- Configuration file values
- Default values
Key Configuration Areas
Database Connectivity
Services use PostgreSQL with sqlx connection pooling. The default pool size is 10 connections, adjustable via --db-max-connections.
Connection strings follow PostgreSQL format:
postgresql://[user[:password]@][host][:port][/database]
Network Settings
Core parameters include:
- Primary RPC:
--node-url
- Simulation RPC:
--simulation-node-url (optional)
- Chain identification:
--chain-id
- Block timing:
--network-block-interval
Contract addresses can be overridden for settlement, balances, and signatures components.
Gas Estimation
Services support multiple estimators (Native, Web3, EthGasStation, GasNow) with configurable price caps.
It is recommended to include “Native” and “Web3” for compatibility across all EVM chains.
Logging System
The system uses tracing-subscriber with flexible EnvFilter syntax:
target[span{field=value}]=level
- Human-readable (default) - For development
- Structured JSON - Preferred for production with log aggregation systems
Runtime Adjustments
Connect to UNIX sockets to modify log filtering without restart:
/tmp/log_filter_override_<program_name>_<pid>.sock
TOML Configuration
Complex settings use TOML files for services like Driver, Autopilot, and Orderbook, supporting structured configuration for solvers, contracts, liquidity, and fee policies.
Network-Specific Presets
| Network | Chain ID | Block Interval |
|---|
| Mainnet | 1 | 12s |
| Gnosis | 100 | 5s |
| Arbitrum | 42161 | 250ms |
| Base | 8453 | 2s |
| Service | Recommended Pool Size |
|---|
| Orderbook | 50 |
| Autopilot | 20 |
| Driver | 15 |
Best Practices
- Store sensitive credentials in environment variables only
- Enable JSON logging for production deployments
- Use appropriate log levels (
info/warn for production, debug/trace for troubleshooting)
- Leverage runtime filtering for production debugging without restarts
- Explicitly set chain IDs to catch configuration errors early
Last modified on March 4, 2026
