[AI discovered CI] dotnet.Tests.dll.18 aborts with core dump on Linux Helix #56465

Build Information

• Current build: dotnet-sdk-public-ci 20261001.9
• Branch: refs/heads/main (ff7814cdd3096e3a9b06c0a37d1d2638733ae892)
• Failing work item: dotnet.Tests.dll.18 in Linux x64 - ubuntu.2204.amd64.open
• Phase: test-execution
• Failure type: process-crash
• Evidence sources: helix-console, process-exit-code, helix-trx
• Failure fingerprint: test-execution|process-crash|dotnet.tests.dll.<n>|aborted-core-dumped-export-_commandexitcode-<n>-xargs--<n>--i-cp-datadisks-disk1-work-c3670a09-w-ae6d094e-uploads-find-datadisks-disk1-work-c3670a09-w-ae6d094e-uploads-..-..-..--na
• Matching prior builds: none identified in the supplied evidence. The related failures in builds 1619713, 1618484, 1618022, 1617646, and 1617544 show materially different test mechanisms and are not counted as matching occurrences.

Failure History

The matching occurrence count is 1: the current build only. The surrounding pipeline sequence was failed (1619713), passed (1618837), then failed (1619854), but the supplied evidence does not establish that the earlier failures had the same crash cause. This is an observed recurrence pattern for the pipeline, not evidence that this specific crash is recurring.

Error Details

The work item reported Aborted (core dumped), exported _commandExitCode=134, and exited with code 134 after 173 of 190 tests passed. The console reported signal 6 while Microsoft.Testing.Platform.IPC.NamedPipeServer.<WaitConnectionAsync>b__27_0() was active, then successfully wrote a crash report and a 290,549,760-byte full dump.

• Helix console log
• TRX result

The supplied artifact list does not include a downloadable dump URL; it only records the dump path in the console evidence.

Root Cause Analysis

• Observed: The dotnet.Tests.dll.18 process terminated with SIGABRT/exit 134 during test execution. The crash occurred while the testing platform was waiting on its named-pipe server, and Helix captured a core dump.
• Assessment: This is an SDK test-execution process crash, not a test assertion or the downstream Monitor Helix Jobs task itself. The evidence establishes the proximate process abort but does not establish whether the trigger is a test-host/runtime defect, an IPC failure, or an environment-specific condition.
• Confidence: Medium for the process-crash characterization; Low for the underlying cause.
• Alternatives / Unknowns: No dump analysis, stack trace, target-branch comparison, or source/change correlation is present. The crash may be isolated to this work-item environment, or may reflect a repeatable test-platform/IPC failure; the supplied evidence cannot distinguish these.

Suggested Investigation

1. Analyze the captured core dump and crash-report JSON for the crashing thread's native and managed stack, especially around the testing-platform named-pipe server.
2. Compare the same dotnet.Tests.dll.18 shard across the next build and a known-good Linux run to determine whether the crash reproduces on the same test set.
3. If reproducible, inspect the affected test-host/IPC path and correlate the crash with the shard's final test and the tested SDK tree; otherwise route as an environment-specific Helix incident.

The generic Monitor Helix Jobs failure is a downstream summary of the failed work item, not the root cause.

Generated by 🕵️ CI Quality Investigator · copilot · gpt56 · 1.53 AIC · ⌖ 0.891 AIC · ⊞ 29.5K · ◷