This runs the entire hardware recommendation. It will run all the tests, produce all the charts and print out the recommendations. Currently it covers Jira L and Jira XL.
- Run the Bamboo plan
- Optionally override the
HWR_PROPS
variable with a value like-Dhwr.jsw.version=7.13.0
- Optionally override the
- Look for results in the Bamboo artifacts
If the Bamboo agent goes offline after 12 hours, rerun the plan. This can take 3-4 reruns.
- Navigate to the repo root
- Set up AWS credentials according to default AWS credentials
- Create a
jira-license.txt
file and fill it with a Jira license - Run
recommendHardware
Gradle task- From terminal:
./gradlew recommendHardware
- Or in short:
./gradlew recHar
- Or from IntelliJ 2019+:
Run anything
(e.g. double tapCtrl
) and typegradle recommendHardware
- From terminal:
- Read the logs and look for results in
build/jpt-workspace
At the beginning of the run, the results from S3 cache (if any) is downloaded and merged with local results. Then the local results are reused. Only the results that are missing will be run. Use this to your advantage. If the build flakes, rerun to just fill in the missing subresults. Naturally, Bamboo does not have any local results so it will always download the entire S3 cache.
The S3 cache requires read/write permissions to the S3 bucket,
so either match AWS creds to the bucket or change the bucket in test source.
The quicksilver-jhwr-cache-ireland
is owned by AWS account jira-server-perf-dev (695067801333)
.
You can browse the shared cached results via S3 GUI. You can also make them public an link them: a cached file.
If you customize the test and want to get a fresh batch of results, change the val cacheKey
in the test source.
Start from the test source and tweak existing knobs and levers or build new ones.
Feel free to chat, ask for help, bounce ideas on the JPT Community Slack.
The CI covers a scaled-down version of the full process. It should be run periodically to prevent breakage caused by code changes or external services.