Run Benchmarks

Run Benchmarks¶

The toolkit ships a lightweight benchmark harness to compare adapters and surface regressions in leasing/health persistence. It avoids network calls by using synthetic health checks so results focus on storage performance.

Quickstart (in-memory)¶

python -m pharox.benchmarks --adapter in-memory --iterations 500 --pool-size 32

This seeds a pool, acquires/releases proxies, then runs synthetic health checks against the same storage. Output includes throughput and hit/miss counts.

Benchmark PostgreSQL¶

Provide a DSN once the Postgres tables exist (or let the script create them):

python -m pharox.benchmarks \
  --adapter postgres \
  --dsn postgresql+psycopg://user:pass@localhost:5432/pharox \
  --iterations 2000 \
  --health-rounds 5

Install the postgres extra first:

pip install "pharox[postgres]"

Tuning knobs¶

• --iterations and --pool-size tune leasing throughput checks.
• --health-proxies, --health-rounds, and --latency-ms adjust synthetic health checks to stress different code paths.
• Results are printed to stdout for easy capture in CI; wire them into your pipeline to guard against regressions.