Concurrency Stress Testing Tools

A small toolkit for stress-testing concurrency-related features in ClickHouse. It runs clickhouse-bench for each query while increasing the number of concurrent clients from 1 to 32. This helps evaluate scheduling- and concurrency-related changes.

Prerequisites

• ClickHouse server running and accessible.
• clickhouse-bench and clickhouse-client available in PATH.
• Bash and Python 3 (for the log-to-JSON converter).
• This directory contains: run.sh, queries.txt, logs2json.py, and index.html.

Testing procedure

1. Start the ClickHouse server and set up any concurrency features you want to test.

2. Prepare queries.txt. The file is semicolon-separated with two columns:

   • Column 1: test duration per concurrency level in seconds.
   • Column 2: SQL query to test.

3. Run the benchmark. Running in the background is recommended:

   Copy

   nohup ./run.sh > run.log 2>&1 &
   

4. After completion, results for each query will be saved into separate Q*.log files. Convert them to JSON:

   Copy

   ./logs2json.py /path/to/workdir/Q*.log > results.json
   

5. Visualize results. Add your result JSON file(s) to index.html:

   jsCopy

   const DATASET_FILES = [
     "./no-concurrency-control.json",
     "./concurrency-control.json"
   ];
   

6. Open index.html in your browser.

Notes

• Default concurrency levels are 1..32. To change them, edit run.sh.
• Ensure queries in queries.txt do not contain ; since it is used as the separator.