[Incident] [incident] aoai-model deployment failure: Smoke Azure Codex and Smoke Azure PI engine steps exit 1

Đang mở
  • Có gì không ổn?
#77,255 0 bình luận 0 reaction 0 người được giao Xem trên GitHub

Maintainer thường phản hồi trong vòng 1 ngày

Chưa có ai nhận issue này.

Đánh giá

Độ khó
4/5
Thời gian dự kiến
3-5 ngày
Mức phù hợp với người mới
43/100
Loại issue
Lỗi
Độ rõ ràng
Khá rõ ràng
Mức độ hoạt động
Sôi nổi
Công nghệ
azure, github-actions, go
Lĩnh vực
ci-cd, infrastructure

Hướng nghiên cứu

Start with the linked failed runs and inspect the Execute Codex CLI and Execute Pi CLI step logs for the driver stderr and exit reason. Then check the aoai-model environment variables, federated credential subject, Entra-authenticated AWF proxy, and Azure OpenAI deployment. Done means both smoke workflows pass when rerun with workflow_dispatch.

Do mô hình lập chỉ mục viết ra từ nội dung của issue.

Mô tả

deployment-failure incident
Summary

Deployments to the aoai-model environment are failing on main at commit c4fffbf1eb. The failing deployments come from two scheduled smoke workflows, Smoke Azure Codex and Smoke Azure PI. In both, the engine step exits with code 1 about 30 seconds after it starts. The same failure has recurred on every scheduled run since 2026-11-03.

Failing runs and deployments
Workflow Run Failing engine step Agent job
Smoke Azure Codex 43334288414 Execute Codex CLI 131586831744
Smoke Azure PI 43334109551 Execute Pi CLI 131586277653

Failed aoai-model deployments: 7753171242, 7753186837, 7753201642 (all failure, 00:08 to 00:09 UTC on 2026-11-13).

Likely root cause
  • Both engine steps fail the same way. The driver exits with code 1 before a terminal safe output is recorded, and the conclusion job then runs Handle agent failure.
  • Shared setup steps (MCP gateway, safe-outputs config, AWF install, container download) all pass. The failure is limited to the engine step. That points at Azure OpenAI inference through the Entra-authenticated AWF proxy (provider: azure, OIDC via AZURE_TENANT_ID, AZURE_CLIENT_ID, AI_GATEWAY_SCOPE).
  • Earlier auto-filed incidents for these workflows (for example #74268 and #76646) were classified driver_crash, which matches.
  • The failures began 2026-11-03 across six different commit SHAs, so no single recent commit explains them. The firewall bump 838595e127 postdates that first failure and is not the trigger.
  • This is an inference. The engine stderr could not be read (see below).
Evidence limitation

gh run view --log-failed returns HTTP 403 (DIFC policy violation) for these runs, so the engine output was not available. The findings above come from job step metadata, check-run annotations, and prior incident records.

Job step timings and annotations
  • Codex agent job 131586831744: Execute Codex CLI 00:07:38 to 00:08:08 UTC, failure. Annotation: Agent execution exited with code 1.
  • Pi agent job 131586277653: Execute Pi CLI 00:06:39 to 00:07:08 UTC, failure. Annotation: Agent execution exited with code 1.
  • Conclusion jobs 131587683233 and 131587268644: Handle agent failure failed.
  • Sibling smoke-copilot-aoai-entra is intermittent: success on 2026-11-09, failure on 2026-11-11 (run 42929934126).
Suggested remediation
  1. Open the failed Execute Codex CLI and Execute Pi CLI steps in the run links above. Capture the driver stderr and exit reason.
  2. Confirm the aoai-model environment variables AZURE_TENANT_ID, AZURE_CLIENT_ID, and AI_GATEWAY_SCOPE are set. Confirm the federated credential subject matches the environment.
  3. Check token acquisition and health of the Entra-authenticated AWF proxy and the Azure OpenAI deployment used by these smokes.
  4. After a fix, re-run both smoke workflows with workflow_dispatch and confirm they pass before closing this incident.

Generated by 🚨 Deployment Incident Monitor · copilot · routed: haiku55 xhigh · ~3 AIC · ⌖ ~1 AIC · ⊞ 2.9K · ◷

  • expires on Nov 19, 2026, 4:15 PM UTC-08:00
Ngôn ngữ chính
Go
Star
5.6k
Fork
625
Merge trung bình
5 giờ 39 phút
Pull request đã merge (30 ngày)
796

Chuẩn bị môi trường

Mở trong Codespaces

Khởi chạy dev container của dự án ngay trên trình duyệt, bằng tài khoản GitHub của bạn.

Bắt đầu từ đâu

  1. Đọc hết issue, rồi đọc hướng dẫn đóng góp của dự án.
  2. Bình luận trên issue rằng bạn sẽ nhận — tránh hai người làm cùng một việc.
  3. Fork repository và làm thay đổi trên một nhánh.
  4. Mở pull request có tham chiếu số hiệu của issue.

Issue khác của github/gh-aw

Tất cả issue của github/gh-aw

Issue tương tự

Thêm issue về Go

Nhận issue mới trong hộp thư của bạn

Bản tóm tắt ngắn những issue GitHub phù hợp với người mới.