demos/jupyter configures NGINX jupyter-proxy to upstream plaintext HTTP on port 80 to atenet-router without mTLS, failing with 502/504

オープン
#3,277 コメント 0 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 1 日以内に返信

まだ誰も着手していません。

評価

難易度
4/5
見積もり時間
3〜5日
初心者へのやさしさ
48/100
issue の種類
バグ
明瞭さ
明確に書かれている
活発さ
活発
技術スタック
go, kubernetes, nginx

調査の方向性

Start with demos/jupyter/jupyter.yaml.tmpl:46-90 and the router Service at manifests/ate-install/atenet-router.yaml:416-433; the README also records the proxy configuration and two stale details. Check how the demo projects pod certificates and the servicedns trust bundle, then reproduce with the issue’s deployment and port-forward commands. Done means the proxy uses the router’s HTTPS service with client authentication, serves JupyterLab, and the README’s cluster and state names are corrected.

索引モデルが issue の本文から書いたものです。

説明

area/demos area/network kind/bug
Summary & Root Cause

Service/atenet-router no longer has a plaintext port 80, and its HTTPS listener requires a client certificate from the pod-identity CA. The jupyter demo's NGINX still does proxy_pass http://atenet-router.ate-system.svc.cluster.local; (implicit port 80), and the Deployment mounts no client cert or trust bundle. Every upstream connection goes to a ClusterIP port that does not exist. With kube-proxy (Kind) that gives an immediate RST and a 502. GKE Dataplane V2 silently drops the SYN, so NGINX waits for its 60s proxy_connect_timeout and returns 504.

Affected Code Locations
  • demos/jupyter/jupyter.yaml.tmpl:46-61 (ConfigMap/jupyter-proxy-config: listen 80; at line 50, proxy_pass http://atenet-router.ate-system.svc.cluster.local; at line 52)
  • demos/jupyter/jupyter.yaml.tmpl:63-90 (Deployment/jupyter-proxy: only the nginx config volume; no podCertificate or clusterTrustBundle volume)
  • demos/jupyter/README.md:19 (claims the Kind cluster is named "ate-demo"), README.md:63-64 (documents the same plaintext proxy_pass), README.md:105 (says STATUS_SUSPENDED)
  • manifests/ate-install/atenet-router.yaml:416-433 (Service/atenet-router exposes only https 443→8443 and connect-tls 8444)
Detailed Reproduction Guide
  1. Deploy the demo and confirm the router ports:

    go run ./cmd/ate-setup deploy demo jupyter
    kubectl get svc -n ate-system atenet-router
    
  2. Hit the proxy:

    kubectl port-forward -n ate-demo-jupyter svc/jupyter-proxy 18081:80 &
    curl -i http://localhost:18081/
    kubectl logs -n ate-demo-jupyter deploy/jupyter-proxy --tail=5
    
Expected vs. Actual Behavior
  • Expected: jupyter-proxy connects to https://atenet-router.ate-system.svc:443, presents a projected podidentity.podcert.ate.dev/identity client certificate, trusts the servicedns bundle, wakes the jupyter-notebook actor, and serves JupyterLab.
  • Actual: Every request fails: 504 Gateway Time-out after about 60s on GKE Dataplane V2, or 502 Bad Gateway where the closed port is actively refused. The README also has a wrong cluster name and a stale state name.
Observed Output & Logs
  $ awk '/proxy_pass|listen/' demos/jupyter/jupyter.yaml.tmpl
          listen 80;
            proxy_pass http://atenet-router.ate-system.svc.cluster.local;
  $ kubectl get svc -n ate-system atenet-router   -> ports https:443->8443, connect-tls:8444->8444   (no 80)
  $ go run ./cmd/ate-setup deploy demo jupyter          # succeeds
  $ kubectl port-forward -n ate-demo-jupyter svc/jupyter-proxy 18081:80 & curl -i http://localhost:18081/
  HTTP/1.1 504 Gateway Time-out          (after ~60s)
  jupyter-proxy: [error] upstream timed out (110: Operation timed out) while connecting to upstream, ... upstream: "http://34.118.234.238:80/"
  HTTP/1.1 504 Gateway Time-out   TIME=60.107376s
  jupyter-proxy: [error] upstream timed out ... upstream: "http://34.118.234.238:80/"
主要言語
Go
スター
5.9k
フォーク
637
平均マージ
2日 1時間
マージ済み PR(30日)
523

環境構築

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

agent-substrate/substrate のほかの issue

agent-substrate/substrate の issue をすべて見る

似ている issue

Go の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。