CI: Take the report merge off the test fan-out, stop the agent pod churn, report split balance - #5104
Open
michaelsembwever wants to merge 2 commits into
Open
CI: Take the report merge off the test fan-out, stop the agent pod churn, report split balance#5104michaelsembwever wants to merge 2 commits into
michaelsembwever wants to merge 2 commits into
Conversation
…urn, report split balance Three faults measured on a Kubernetes clone of this pipeline. Jenkinsfile: - generateTestReports moves from cassandra-medium, the label of the ~460 test splits it queued behind, to cassandra-report: 109 minutes of a 163 minute build, for ~3 minutes of work - one ant junitreport per target, three at a time under an explicit -Xmx - size every `xargs -P` and `xz -T` from the container's cgroup quota. nproc answers for the node: 8 against a 2 cpu limit - record each cell's duration and longest suites, and archive them - test-burn takes 5 splits, which separates its three long classes - remove testDataPublishers as they were oom-ing ci-cassandra.a.o's controller https://the-asf.slack.com/archives/CK23JSY2K/p1787841152008959?thread_ts=1787836602.176469&cid=CK23JSY2K , and it was noise because of jenkinsci/test-stability-plugin#11 (we use butler instead) jenkins-deployment.yaml: - slaveConnectTimeout 600 and waitForPodSec 900, the two deadlines on one provisioning attempt. 30 seconds met only a warm node, and churned 1,110 pods in one 24 minute window with 0 agents connected and 483 `KubernetesProvisioningLimits ... went below zero` warnings - idleMinutes 5, a pod holding a whole node - the four instanceCaps to the demand a pre-commit run measured; containerCap to their sum, so no pool's share is decided by which holds its agents longest, and maxRequestsPerHostStr with it - add the agent-dind-report template, on a node pool of its own cell_balance.py: - new. Per target, how evenly its splits divided and what the worst cell ran, to the console and ci_summary.html. Splits stay round-robin, so the lever is the split count - guarded in generate-ci-summary.sh, never failing a build run-ci-test.py: - the fixture's node group maximums follow the raised caps patch by Mick Semb Wever; reviewed by xxx for CASSANDRA-##### Assisted-by: Claude Code:claude-opus-5
driftx
requested changes
Sep 3, 2026
| // - cassandra-small + cassandra-${arch}-small : 1 cpu, 1GB ram (alias for above but for any arch) | ||
| // - cassandra-medium + cassandra-${arch}-medium : 3 cpu, 5GB ram | ||
| // - cassandra-large + cassandra-${arch}-large : 7 cpu, 16GB ram | ||
| // - cassandra-report + cassandra-${arch}-report : 3 cpu, 5GB ram |
Contributor
There was a problem hiding this comment.
I don't think this comment matches agent-dind-report?
Member
Author
There was a problem hiding this comment.
thanks, addressed
| CASSANDRA_DOCKER_ANT_OPTS="-Dbuild.test.output.dir=build/test/output/\${target} -Dbuild.test.report.dir=build/test/reports/\${target}" | ||
| find build/test/output -type f -name "*.xml.xz" -print0 | xargs -0 -r -n64 -P"\$(cpus)" xz -f --decompress | ||
|
|
||
| # One ant junitreport per test target, three at a time. |
Contributor
There was a problem hiding this comment.
"up to three at a time" I think, report_jobs is effectively min(cpus(), 3), and cpus() here is the jnlp container's limit (2), so this pod runs two.
Member
Author
There was a problem hiding this comment.
thanks, addressed
| """ | ||
| } else { | ||
| copyArtifacts filter: 'test/**/TEST-*.xml.xz,test/**/cqlshlib*.xml.xz,test/**/nosetests*.xml.xz,test/**/jmh-result.json', fingerprintArtifacts: true, projectName: env.JOB_NAME, selector: specific(env.BUILD_NUMBER), target: "build/", optional: true | ||
| copyArtifacts filter: 'test/cell-times/**,test/**/TEST-*.xml.xz,test/**/cqlshlib*.xml.xz,test/**/nosetests*.xml.xz,test/**/jmh-result.json', fingerprintArtifacts: true, projectName: env.JOB_NAME, selector: specific(env.BUILD_NUMBER), target: "build/", optional: true |
Contributor
There was a problem hiding this comment.
need to copy cell-times in the above isCanonical too
Member
Author
There was a problem hiding this comment.
thanks, addressed
| if args.output and os.path.isfile(args.output): | ||
| try: | ||
| with open(args.output, "a", encoding="utf-8") as handle: | ||
| handle.write(html_report(rows, args.budget_margin)) |
Contributor
There was a problem hiding this comment.
ci_parser.py closed the body and html tags before this, so it won't be valid html.
Member
Author
There was a problem hiding this comment.
thanks, addressed
driftx
approved these changes
Sep 4, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Three faults measured on a Kubernetes clone of this pipeline.
Jenkinsfile:
xargs -Pandxz -Tfrom the container's cgroup quota. nproc answers for the node: 8 against a 2 cpu limitjunit()call when a build publishes results more than once jenkinsci/test-stability-plugin#11 (we use butler instead)jenkins-deployment.yaml:
KubernetesProvisioningLimits ... went below zerowarningscell_balance.py:
run-ci-test.py: