Solving Blockchain Load Testing with Locust: Leveraging Python-Based Scenarios and Master-Worker Architecture

Blockchain is difficult to load test with existing tools due to nonce and signature handling. Locust solves this problem through flexible Python-based scenario writing and integration with the web3 module.

Verifying Request Rate Limits

A server allowing 100 requests per second was configured in Golang, and 110 virtual users were generated with Locust to confirm 429 Too Many Requests responses. This verified that the server's Rate Limiting logic was working correctly.

Large-Scale Traffic Generation Architecture

To overcome the computing resource limits of a single node, a Master-Worker architecture was applied. Docker Compose was used to configure 1 master and 3 worker nodes to generate distributed load, and the resource status of each container was monitored.

Simulating Users with Unique Wallets

For blockchain transaction testing, each user was implemented to be assigned a unique wallet upon initialization. In the on_start method, each user was given a unique name, and by integrating with a Go server that tracks click counts, traffic was precisely controlled on a per-member basis.