Skip to content

Make threat-detect binary install step continue-on-error in warn mode - #52400

Draft
pelikhan with Copilot wants to merge 2 commits into
mainfrom
copilot/aw-failures-harden-binary-installers
Draft

Make threat-detect binary install step continue-on-error in warn mode#52400
pelikhan with Copilot wants to merge 2 commits into
mainfrom
copilot/aw-failures-harden-binary-installers

Conversation

Copilot AI commented Aug 13, 2026

Copy link
Copy Markdown
Contributor

A GitHub Releases CDN blip caused the evals/detection job installers for the AWF firewall and threat-detect binaries to fail across multiple PR-gate workflows within a 25-minute window. The threat-detect binary download already had retry-with-backoff (curl --retry 5 --retry-delay 10 --retry-max-time 180), but a hard failure after retries still marked the whole detection job as failure — even though the job's own logic (GH_AW_DETECTION_CONTINUE_ON_ERROR) already treats a missing threat-detect binary as non-fatal in warn mode.

Compiler change

  • buildInstallThreatDetectStep now takes *WorkflowData and emits continue-on-error: true (or the configured expression) for the "Install threat-detect binary" step, reusing the same continue-on-error resolution already applied to the detection conclusion/analysis steps:
- name: Install threat-detect binary
  if: always() && steps.detection_guard.outputs.run_detection == 'true'
  continue-on-error: true
  run: |
    bash "${RUNNER_TEMP}/gh-aw/actions/install_threat_detect_binary.sh" latest

This keeps the install step's failure semantics consistent with the job's existing tolerance policy: in warn mode a transient download failure no longer flips the job conclusion to failure; in strict mode (continue-on-error: false) the step still fails the job as before.

Test coverage

  • threat_detection_isolation_test.go: added an assertion in TestExternalDetectorPath verifying the install step carries continue-on-error: true in the default warn mode.

Generated workflows

  • Recompiled all .lock.yml files; 104 workflows using features: gh-aw-detection: true pick up the new continue-on-error: true line on this step.

Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>
Copilot AI changed the title [WIP] Add retry-with-backoff to AWF/threat-detect binary installers Make threat-detect binary install step continue-on-error in warn mode Aug 13, 2026
Copilot AI requested a review from pelikhan August 13, 2026 03:20
@github-actions

Copy link
Copy Markdown
Contributor

PR Triage

Category: chore | Risk: low | Priority: low | Score: 31/100

Score breakdown: impact 15/50, urgency 8/30, quality 8/20

CI: unknown (draft)

Recommended action: defer

CI hardening chore, still draft/blocked.

Automated triage — run 52400

Generated by 🔧 PR Triage Agent · auto · 67.2 AIC · ⌖ 2.77 AIC · ⊞ 7.8K ·

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[aw-failures] [P0] Harden AWF/threat-detect binary installers against transient GitHub Releases connection failures

2 participants