You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[AUTOCUT] Integration Test Failed for sql-3.7.0 #4
Check out test report manifest linked above for steps to reproduce, cluster and integration test failure logs. For additional information checkout the wiki and OpenSearch Metrics Dashboard.
[Needs info] Integ-test failure is Maven Central HTTP 429 throttling, not a SQL code bug — no fix to apply
The linked CI run failed during Gradle buildscript classpath resolution before any SQL test executed. Gradle could not download plugin POMs from repo.maven.apache.org because the host returned HTTP 429 (Too Many Requests). The same 429 errors caused the alerting and ml-commons components in the same run to fail in their configuration phase, which indicates this is a transient, environment-wide rate-limiting event rather than a SQL-side defect. The ci.opensearch.org/maven2/ mirror that exists specifically to absorb this kind of throttling is already on main (commit 7b4ebcfa9, PR opensearch-project#5478), which is why I'm not opening a code-change PR.
CI distribution under test: 3.7.0 build 11969 RC1, integ-test run 11215
Analysis
The integ-test stderr for the SQL component is a Gradle configuration-phase failure, not a test-execution failure. The exception fires before any task runs:
Could not resolve com.netflix.nebula.ospackage-base:com.netflix.nebula.ospackage-base.gradle.plugin:12.0.0
Could not resolve io.freefair.lombok:io.freefair.lombok.gradle.plugin:8.14
Could not resolve com.diffplug.spotless:com.diffplug.spotless.gradle.plugin:8.1.0
Each underlying Could not HEAD … line ends with Received status code 429 from server: Too Many Requests against repo.maven.apache.org. Build totals: BUILD FAILED in 3s — too fast for any test to run.
The same repo.maven.apache.org 429 pattern appears in the alerting and ml-commons stderrs from the same run (org.jetbrains.kotlin:kotlin-gradle-plugin:2.2.0, com.dorongold.task-tree, com.netflix.nebula.ospackage, etc.). Three independent component repos with disjoint plugin sets failing the same way at configuration time is the signature of upstream-host throttling, not a per-repo code defect.
CI distribution build 11969 was produced before that mitigation landed, which is the most likely reason the failing run still hit Maven Central directly. A re-cut on a build that picks up main @ d099d3eff (or anything ≥ 7b4ebcfa9) should resolve plugins from the ci.opensearch.org mirror first and avoid the 429 path entirely.
Evidence
Excerpted from the CI test report manifest test-report.yml (multiple components, all failed_test: Test Result Files List Not Available, which is the build-system signal for "tests never started"):
FAILURE: Build failed with an exception.
* What went wrong:
A problem occurred configuring root project 'opensearch-sql'.
> Could not resolve all artifacts for configuration 'classpath'.
> Could not resolve com.netflix.nebula.ospackage-base:com.netflix.nebula.ospackage-base.gradle.plugin:12.0.0.
...
> Could not HEAD '...nebula.ospackage-base.gradle.plugin-12.0.0.pom'. Received status code 429 from server: Too Many Requests
> Could not resolve io.freefair.lombok:io.freefair.lombok.gradle.plugin:8.14.
...
> Received status code 429 from server: Too Many Requests
> Could not resolve com.diffplug.spotless:com.diffplug.spotless.gradle.plugin:8.1.0.
...
> Received status code 429 from server: Too Many Requests
> There are 21 more failures with identical causes.
[truncated, full file is 33 lines]
BUILD FAILED in 3s
FAILURE: Build failed with an exception.
* What went wrong:
A problem occurred configuring root project 'opensearch-alerting'.
> Could not resolve all artifacts for configuration 'classpath'.
> Could not resolve org.jetbrains.kotlin:kotlin-gradle-plugin:2.2.0.
...
> Received status code 429 from server: Too Many Requests
> Could not resolve com.netflix.nebula.ospackage:com.netflix.nebula.ospackage.gradle.plugin:12.0.0.
...
> Received status code 429 from server: Too Many Requests
Re-run the integ-test pipeline on a distribution build that includes commit 7b4ebcfa9 (i.e., re-cut RC2 from current main, or a fresh integ-test job that re-clones main). The ci.opensearch.org/maven2/ mirror should keep Gradle off the throttled path.
If a re-run on a post-7b4ebcfa9 distribution still fails with repo.maven.apache.org 429s, the mitigation is being bypassed and this becomes a real regression — please reopen with the new test-report URL and I'll investigate which resolution path is still falling through to Maven Central.
Because the same 429 pattern hit alerting and ml-commons, please consider tracking the cross-component infrastructure side in opensearch-project/opensearch-build (the test-runner repo) so that other plugins also get a mirror-first plugin resolution policy. That is out of scope for the SQL repo.
Confirm with me whether you want this issue closed as a transient CI infrastructure failure, or kept open as the tracker for the RC2 re-cut. I'm declining to write a code-change PR because there is no SQL source change that would fix what failed here.
Activity
RyanL1997 commentedon May 28, 2026
@ryan-gh-bot /fix
ryan-gh-bot commentedon May 28, 2026
Working on
/fix(invoked by @RyanL1997) — posting result when done.ryan-gh-bot commentedon May 28, 2026
[Needs info] Integ-test failure is Maven Central HTTP 429 throttling, not a SQL code bug — no fix to apply
The linked CI run failed during Gradle buildscript classpath resolution before any SQL test executed. Gradle could not download plugin POMs from
repo.maven.apache.orgbecause the host returned HTTP 429 (Too Many Requests). The same 429 errors caused the alerting and ml-commons components in the same run to fail in their configuration phase, which indicates this is a transient, environment-wide rate-limiting event rather than a SQL-side defect. Theci.opensearch.org/maven2/mirror that exists specifically to absorb this kind of throttling is already onmain(commit7b4ebcfa9, PR opensearch-project#5478), which is why I'm not opening a code-change PR.Environment
sql(opensearch-project/sql, targetRyanL1997/sql)target/main @ d099d3effAnalysis
The integ-test stderr for the SQL component is a Gradle configuration-phase failure, not a test-execution failure. The exception fires before any task runs:
Could not resolve com.netflix.nebula.ospackage-base:com.netflix.nebula.ospackage-base.gradle.plugin:12.0.0Could not resolve io.freefair.lombok:io.freefair.lombok.gradle.plugin:8.14Could not resolve com.diffplug.spotless:com.diffplug.spotless.gradle.plugin:8.1.0Each underlying
Could not HEAD …line ends withReceived status code 429 from server: Too Many Requestsagainstrepo.maven.apache.org. Build totals:BUILD FAILED in 3s— too fast for any test to run.The same
repo.maven.apache.org429 pattern appears in the alerting and ml-commons stderrs from the same run (org.jetbrains.kotlin:kotlin-gradle-plugin:2.2.0,com.dorongold.task-tree,com.netflix.nebula.ospackage, etc.). Three independent component repos with disjoint plugin sets failing the same way at configuration time is the signature of upstream-host throttling, not a per-repo code defect.The plugin classpath in
build.gradleL80-L86 declares the three plugins named in the failure. Plugin-id resolution is governed by thepluginManagementblock insettings.gradleL6-L12, and thebuildscriptclasspath also goes throughbuild.gradleL68-L73. Both already listhttps://ci.opensearch.org/maven2/ahead ofmavenCentral(), exactly to bypass Maven Central throttling. That mitigation landed on mainline yesterday in commit7b4ebcfa9(PR Add ci.opensearch.org maven2 mirror to avoid throttling (sql) opensearch-project/sql#5478, May 27 2026).CI distribution build 11969 was produced before that mitigation landed, which is the most likely reason the failing run still hit Maven Central directly. A re-cut on a build that picks up
main @ d099d3eff(or anything ≥7b4ebcfa9) should resolve plugins from theci.opensearch.orgmirror first and avoid the 429 path entirely.Evidence
Excerpted from the CI test report manifest
test-report.yml(multiple components, allfailed_test: Test Result Files List Not Available, which is the build-system signal for "tests never started"):SQL
without-securitystderr (source):Alerting
without-securitystderr (source):Existing mitigation already on
main(build.gradleL66-L74):buildscript { ... repositories { mavenLocal() maven { url "https://ci.opensearch.org/maven2/" } mavenCentral() maven { url "https://ci.opensearch.org/ci/dbc/snapshots/maven/" } } ... }And in
settings.gradleL6-L12:pluginManagement { repositories { maven { url "https://ci.opensearch.org/maven2/" } gradlePluginPortal() mavenCentral() } }Next steps
7b4ebcfa9(i.e., re-cut RC2 from currentmain, or a fresh integ-test job that re-clonesmain). Theci.opensearch.org/maven2/mirror should keep Gradle off the throttled path.7b4ebcfa9distribution still fails withrepo.maven.apache.org429s, the mitigation is being bypassed and this becomes a real regression — please reopen with the new test-report URL and I'll investigate which resolution path is still falling through to Maven Central.opensearch-project/opensearch-build(the test-runner repo) so that other plugins also get a mirror-first plugin resolution policy. That is out of scope for the SQL repo.Automated response from @ryan-gh-bot (maintainer-triggered agent). About this agent.