mirror of
https://github.com/apache/camel-performance-tests.git
synced 2026-11-06 12:48:34 +00:00
Apache Camel Performances Tests Repository
https://camel.apache.org
- Python 48.4%
- Java 42.2%
- Shell 8.4%
- XSLT 0.8%
- JavaScript 0.1%
- Other 0.1%
| Filename | Latest commit message | Latest commit date |
|---|---|---|
9175ca3aec
On Nov 6 a reference pass of test-citrus-orders stopped chaosdemo, a Camel app that another session ran on this machine: run_test stopped every Camel app that started while camel test run ran. run-stepwise-ladder.sh also ran camel stop with no name after each example and camel infra stop for every service of the steps folder after a run, and infra_start stopped a service before it started its own. The apps of the harness are now the ones that run from its folders: the runtime.directory of the app's status file, as camel restart takes it, or its working directory. That is the dev app, the peer, the integrations a Citrus test starts, and the apps the model starts with camel_run under any name. run_test stops those the test left running, the harness stops what still runs of them when an example ends, and the runners do the same as a safety net for an interrupted run. Another session's apps run from its own folders and stay. The harness records the PID of each infra service it starts, in stepwise/infra-started.json, and stops only those: before the next pass of an example and when the run is over. A service that another session runs already is used as it is and left running, with a line in run.log. run-kamelet-check.sh gets the same two changes. agent_local.py, the one-shot harness, is unchanged. Verified with reference and negative passes of test-citrus-orders, connect-http-client and connect-service-jms-request-reply, with a Camel app from another folder running beside them: it ran through all passes, and after each pass no app or infra service of the harness was left. An artemis started outside the harness was used as it is and not stopped. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_016Y4dDi5iQNgtHrDdz2NpwF |
||
| .github | ||
| ai-benchmark | ||
| profiling | ||
| tests | ||
| .asf.yaml | ||
| .gitignore | ||
| LICENSE.txt | ||
| NOTICE.txt | ||
| pom.xml | ||
| README.md | ||
Apache Camel Performance Tests
This project provides performance tests for Apache Camel.
Running the Performance Tests
Camel JMH
The performance tests located in the camel-jmh module can be run using the following command:
mvn -l camel-${version}-jmh-test.log -DargLine="-XX:+UseNUMA -Xmx4G -Xms4G -server" -Dcamel.version=${version} -Pjmh -Dtest=${TESTS} clean test
Where:
${version}is the Apache Camel version to use.${TESTS}is a comma-separated list of tests to run (e.g.,AggregatorTest,ContentBasedRouterBodyTest).
After running the tests, the results are stored in JMH JSON files (i.e.: <test name>.jmh.json).
AI benchmark
The ai-benchmark directory holds a harness that measures how well an AI model builds and edits Camel integrations
through the Camel MCP server, scored by what runs. See ai-benchmark/README.md and the
blog post https://camel.apache.org/blog/2026/09/camel-local-model-benchmark/ for the first series.