Disable CDS in nested JVMs of CI Visibility smoke tests on Linux arm64

[daniel-mohedano]

What Does This Do

CI Visibility smoke tests now disable CDS (-Xshare:off) on Linux arm64 for every JVM they start, including nested ones:

• JVMs started with the agent arguments from CiVisibilitySmokeTest.buildJvmArguments get -Xshare:off directly. This covers the Gradle TestKit daemon (through org.gradle.jvmargs), Maven, and the JUnit Console launcher.
• Nested JVMs get -Xshare:off through JAVA_TOOL_OPTIONS in the environment of the started processes:
  • GradleLauncherSmokeTest: Gradle wrapper and launcher, and the daemon it starts
  • GradleDaemonSmokeTest: test workers forked by the TestKit daemon
  • MavenSmokeTest: Surefire forks

The change does nothing on other platforms.

Motivation

The other smoke tests already disable CDS on Linux arm64 to avoid JVM crashes (see AbstractSmokeTest and AbstractSmokeApp). The CI Visibility smoke tests did not, and they start several levels of JVMs.

On master, these tests had about 109 failures over the last 3 weeks from JVM crashes on arm64:

• GradleLauncherSmokeTest: 102 (launcher exits with code 134)
• GradleDaemonSmokeTest: 4 (SIGSEGV in ClassLoaderData::add_handle)
• MavenSmokeTest: 3

That makes this the main source of flakiness for these suites.

Additional Notes

I could not reproduce the crash locally. It happens in about 0.7% of the Gradle launcher runs on arm64 CI. CI runs on this branch should confirm whether it fixes the issue.

test-environment-trigger: skip

[cit-pr-commenter-54b7da]

CI Visibility Test Environment - sbt-scalatest

Job Status: 🟢 success

Scenario	This PR (%)	7d median	Δ 7d	30d median	Δ 30d	runs (7d/30d)
agent	57.59	57.69	$\color{green}{\blacktriangledown}$ -0.10	56.55	$\color{red}{\blacktriangle}$ +1.04	70/316
agentEvpProxy	57.04	n/a	n/a	n/a	n/a	-

Baseline: median of @test.tracer_overhead on main (gitlab) over the last 7/30 days, per OSS project & scenario. Δ = this PR − baseline median; red ▲ = more overhead, green ▽ = less overhead than baseline.

Tests run to verify that CI Visibility behavior has not regressed in the current PR.

[datadog-datadog-us1-prod]

🎯 Code Coverage (details)
• Patch Coverage: 100.00%
• Overall Coverage: 59.71% (-0.10%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 86ea008 | Docs | Give us feedback!}

[cit-pr-commenter-54b7da]

CI Visibility Test Environment - nebula-release-plugin

Job Status: 🟢 success

Scenario	This PR (%)	7d median	Δ 7d	30d median	Δ 30d	runs (7d/30d)
agent	37.37	37.15	$\color{red}{\blacktriangle}$ +0.22	37.15	$\color{red}{\blacktriangle}$ +0.22	36/172
agentless	37.58	37.15	$\color{red}{\blacktriangle}$ +0.43	36.42	$\color{red}{\blacktriangle}$ +1.16	36/145
agentlessCodeCoverage	46.19	45.38	$\color{red}{\blacktriangle}$ +0.81	45.38	$\color{red}{\blacktriangle}$ +0.81	36/144
agentlessLineCoverage	57.19	56.55	$\color{red}{\blacktriangle}$ +0.64	56.55	$\color{red}{\blacktriangle}$ +0.64	36/144

Baseline: median of @test.tracer_overhead on main (gitlab) over the last 7/30 days, per OSS project & scenario. Δ = this PR − baseline median; red ▲ = more overhead, green ▽ = less overhead than baseline.

Tests run to verify that CI Visibility behavior has not regressed in the current PR.

[cit-pr-commenter-54b7da]

CI Visibility Test Environment - heliboard

Job Status: 🟢 success

Scenario	This PR (%)	7d median	Δ 7d	30d median	Δ 30d	runs (7d/30d)
agent	9.73	9.92	$\color{green}{\blacktriangledown}$ -0.19	9.92	$\color{green}{\blacktriangledown}$ -0.19	35/141

Baseline: median of @test.tracer_overhead on main (gitlab) over the last 7/30 days, per OSS project & scenario. Δ = this PR − baseline median; red ▲ = more overhead, green ▽ = less overhead than baseline.

Tests run to verify that CI Visibility behavior has not regressed in the current PR.

[cit-pr-commenter-54b7da]

CI Visibility Test Environment - pass4s

Job Status: 🟢 success

Scenario	This PR (%)	7d median	Δ 7d	30d median	Δ 30d	runs (7d/30d)
agent	10.93	10.75	$\color{red}{\blacktriangle}$ +0.18	10.13	$\color{red}{\blacktriangle}$ +0.80	37/167
agentless	8.43	9.92	$\color{green}{\blacktriangledown}$ -1.49	9.92	$\color{green}{\blacktriangledown}$ -1.49	36/137
agentlessCodeCoverage	15.08	18.08	$\color{green}{\blacktriangledown}$ -3.00	17.73	$\color{green}{\blacktriangledown}$ -2.65	31/124

Baseline: median of @test.tracer_overhead on main (gitlab) over the last 7/30 days, per OSS project & scenario. Δ = this PR − baseline median; red ▲ = more overhead, green ▽ = less overhead than baseline.

Tests run to verify that CI Visibility behavior has not regressed in the current PR.

[cit-pr-commenter-54b7da]

CI Visibility Test Environment - netflix-zuul

Job Status: 🟢 success

Scenario	This PR (%)	7d median	Δ 7d	30d median	Δ 30d	runs (7d/30d)
agent	90.59	91.39	$\color{green}{\blacktriangledown}$ -0.80	89.58	$\color{red}{\blacktriangle}$ +1.01	36/173
agentless	82.88	82.69	$\color{red}{\blacktriangle}$ +0.19	82.69	$\color{red}{\blacktriangle}$ +0.19	36/143
agentlessCodeCoverage	98.51	97.04	$\color{red}{\blacktriangle}$ +1.47	97.04	$\color{red}{\blacktriangle}$ +1.47	36/143
agentlessLineCoverage	114.75	113.88	$\color{red}{\blacktriangle}$ +0.87	113.88	$\color{red}{\blacktriangle}$ +0.87	36/144

Baseline: median of @test.tracer_overhead on main (gitlab) over the last 7/30 days, per OSS project & scenario. Δ = this PR − baseline median; red ▲ = more overhead, green ▽ = less overhead than baseline.

Tests run to verify that CI Visibility behavior has not regressed in the current PR.

[cit-pr-commenter-54b7da]

CI Visibility Test Environment - reactive-streams-jvm

Job Status: 🟢 success

Scenario	This PR (%)	7d median	Δ 7d	30d median	Δ 30d	runs (7d/30d)
agent	22.77	22.54	$\color{red}{\blacktriangle}$ +0.23	22.09	$\color{red}{\blacktriangle}$ +0.68	37/172
agentless	19.93	19.59	$\color{red}{\blacktriangle}$ +0.34	19.59	$\color{red}{\blacktriangle}$ +0.34	36/139
agentlessCodeCoverage	20.11	20.39	$\color{green}{\blacktriangledown}$ -0.28	19.99	$\color{red}{\blacktriangle}$ +0.12	36/139
agentlessLineCoverage	27.32	27.52	$\color{green}{\blacktriangledown}$ -0.20	27.52	$\color{green}{\blacktriangledown}$ -0.20	22/94

Baseline: median of @test.tracer_overhead on main (gitlab) over the last 7/30 days, per OSS project & scenario. Δ = this PR − baseline median; red ▲ = more overhead, green ▽ = less overhead than baseline.

Tests run to verify that CI Visibility behavior has not regressed in the current PR.

[dd-octo-sts]

🟢 Java Benchmark SLOs — All performance SLOs passed

Suite	Status
Startup	🟢 pass

SLO thresholds are defined here based on automatically generated metrics. A warning is raised when results are within 5% of the threshold.

PR vs. master results

Scenario	Candidate	master	Δ (95% CI of mean)
startup:insecure-bank:iast:Agent	13.96 s	13.80 s	[+0.4%; +2.0%] (maybe worse)
startup:insecure-bank:tracing:Agent	12.87 s	12.91 s	[-0.8%; +0.2%] (no difference)
startup:petclinic:appsec:Agent	17.14 s	16.25 s	[+1.1%; +9.9%] (significantly worse)
startup:petclinic:iast:Agent	16.89 s	17.07 s	[-1.8%; -0.3%] (maybe better)
startup:petclinic:profiling:Agent	16.87 s	16.82 s	[-0.7%; +1.3%] (no difference)
startup:petclinic:sca:Agent	16.98 s	17.00 s	[-1.0%; +0.7%] (no difference)
startup:petclinic:tracing:Agent	16.09 s	16.09 s	[-0.7%; +0.8%] (no difference)

Commit: 86ea008d · CI Pipeline · Benchmarking Platform UI

---

Load and DaCapo benchmarks can be triggered manually in the GitLab pipeline. Results will appear in the Benchmarking Platform UI after completion.

[cit-pr-commenter-54b7da]

CI Visibility Test Environment - sonar-kotlin

Job Status: 🟢 success

Scenario	This PR (%)	7d median	Δ 7d	30d median	Δ 30d	runs (7d/30d)
agent	13.50	13.67	$\color{green}{\blacktriangledown}$ -0.17	13.40	$\color{red}{\blacktriangle}$ +0.10	36/171
agentless	13.05	12.12	$\color{red}{\blacktriangle}$ +0.93	12.12	$\color{red}{\blacktriangle}$ +0.93	36/141
agentlessCodeCoverage	16.64	15.72	$\color{red}{\blacktriangle}$ +0.92	15.41	$\color{red}{\blacktriangle}$ +1.23	36/141
agentlessLineCoverage	18.49	17.38	$\color{red}{\blacktriangle}$ +1.11	17.38	$\color{red}{\blacktriangle}$ +1.11	35/141

Baseline: median of @test.tracer_overhead on main (gitlab) over the last 7/30 days, per OSS project & scenario. Δ = this PR − baseline median; red ▲ = more overhead, green ▽ = less overhead than baseline.

Tests run to verify that CI Visibility behavior has not regressed in the current PR.

[cit-pr-commenter-54b7da]

CI Visibility Test Environment - jolokia

Job Status: 🟢 success

Scenario	This PR (%)	7d median	Δ 7d	30d median	Δ 30d	runs (7d/30d)
agent	96.72	95.12	$\color{red}{\blacktriangle}$ +1.60	95.12	$\color{red}{\blacktriangle}$ +1.60	37/172
agentless	92.33	91.39	$\color{red}{\blacktriangle}$ +0.94	89.58	$\color{red}{\blacktriangle}$ +2.75	36/143
agentlessCodeCoverage	101.59	101.00	$\color{red}{\blacktriangle}$ +0.59	99.00	$\color{red}{\blacktriangle}$ +2.59	36/143
agentlessLineCoverage	104.23	103.04	$\color{red}{\blacktriangle}$ +1.19	101.00	$\color{red}{\blacktriangle}$ +3.23	42/152

Baseline: median of @test.tracer_overhead on main (gitlab) over the last 7/30 days, per OSS project & scenario. Δ = this PR − baseline median; red ▲ = more overhead, green ▽ = less overhead than baseline.

Tests run to verify that CI Visibility behavior has not regressed in the current PR.

[cit-pr-commenter-54b7da]

CI Visibility Test Environment - okhttp

Job Status: 🟢 success

Scenario	This PR (%)	7d median	Δ 7d	30d median	Δ 30d	runs (7d/30d)
agent	18.14	18.82	$\color{green}{\blacktriangledown}$ -0.68	19.99	$\color{green}{\blacktriangledown}$ -1.85	36/169
agentless	18.67	18.45	$\color{red}{\blacktriangle}$ +0.22	19.20	$\color{green}{\blacktriangledown}$ -0.53	37/145
agentlessCodeCoverage	21.55	21.65	$\color{green}{\blacktriangledown}$ -0.10	22.99	$\color{green}{\blacktriangledown}$ -1.44	38/147
agentlessLineCoverage	39.95	37.91	$\color{red}{\blacktriangle}$ +2.04	39.45	$\color{red}{\blacktriangle}$ +0.50	35/142

Baseline: median of @test.tracer_overhead on main (gitlab) over the last 7/30 days, per OSS project & scenario. Δ = this PR − baseline median; red ▲ = more overhead, green ▽ = less overhead than baseline.

Tests run to verify that CI Visibility behavior has not regressed in the current PR.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-16T10:54:29.730104Z	326d588	Draft marked ready
🔒 Security Review	✅ Completed 2026-10-16T10:54:44.025774Z	326d588	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[cit-pr-commenter-54b7da]

CI Visibility Test Environment - spring_boot

Job Status: 🟢 success

Scenario	This PR (%)	7d median	Δ 7d	30d median	Δ 30d	runs (7d/30d)
agent	17.22	16.36	$\color{red}{\blacktriangle}$ +0.86	16.36	$\color{red}{\blacktriangle}$ +0.86	36/168
agentless	10.60	9.92	$\color{red}{\blacktriangle}$ +0.68	9.92	$\color{red}{\blacktriangle}$ +0.68	36/140
agentlessCodeCoverage	14.44	13.67	$\color{red}{\blacktriangle}$ +0.77	13.67	$\color{red}{\blacktriangle}$ +0.77	36/142
agentlessLineCoverage	23.40	22.54	$\color{red}{\blacktriangle}$ +0.86	22.54	$\color{red}{\blacktriangle}$ +0.86	35/140

Baseline: median of @test.tracer_overhead on main (gitlab) over the last 7/30 days, per OSS project & scenario. Δ = this PR − baseline median; red ▲ = more overhead, green ▽ = less overhead than baseline.

Tests run to verify that CI Visibility behavior has not regressed in the current PR.

[datadog-datadog-us1-prod]

The TestKit daemon still starts with CDS enabled because its build environment is applied after JVM startup. Its daemon JVM arguments also need the Linux arm64 workaround.

_{Open Bits AI session}

_{🤖 Bits Code Review · Commit 326d588}

[cit-pr-commenter-54b7da]

CI Visibility Test Environment - sonar-java

Job Status: 🟢 success

Scenario	This PR (%)	7d median	Δ 7d	30d median	Δ 30d	runs (7d/30d)
agent	31.83	17.38	$\color{red}{\blacktriangle}$ +14.45	15.41	$\color{red}{\blacktriangle}$ +16.42	36/170
agentless	32.88	13.94	$\color{red}{\blacktriangle}$ +18.94	13.67	$\color{red}{\blacktriangle}$ +19.21	36/142
agentlessCodeCoverage	76.67	76.33	$\color{red}{\blacktriangle}$ +0.34	79.45	$\color{green}{\blacktriangledown}$ -2.78	36/142
agentlessLineCoverage	97.50	118.53	$\color{green}{\blacktriangledown}$ -21.03	123.36	$\color{green}{\blacktriangledown}$ -25.86	36/145

Baseline: median of @test.tracer_overhead on main (gitlab) over the last 7/30 days, per OSS project & scenario. Δ = this PR − baseline median; red ▲ = more overhead, green ▽ = less overhead than baseline.

Tests run to verify that CI Visibility behavior has not regressed in the current PR.