chore: remove local testing scripts, docker demo env, and screenshots from repo

- gitignore deploy/ and gui-test-screenshots/ (local dev artifacts, not for repo)
- README: remove deploy section and build path reference
- Files kept on disk for local use, just untracked

.gitignore

@@ -23,11 +23,10 @@ data/
 
 # docker demo staging (frp binaries)
 frp-bin/
-deploy/frp-bin/
 
-# demo runtime data + compiled artifacts (regenerable)
-deploy/data/
-deploy/artifacts/
+# local testing scripts, docker demo env, and screenshots (not for repo)
+deploy/
+gui-test-screenshots/
 
 # vendored deps (regenerable; keep out of git)
 vendor/

README.md

@@ -57,7 +57,7 @@ rm -rf internal/httpapi/dist && cp -r web/dist internal/httpapi/dist
 一键全量构建：
 
 ```bash
-cd web && npm run build && cd .. && cp -r web/dist/* internal/httpapi/dist/ && go build -o deploy/artifacts/webui4frpc ./cmd/webui4frpc
+cd web && npm run build && cd .. && cp -r web/dist/* internal/httpapi/dist/ && go build -o webui4frpc ./cmd/webui4frpc
 ```
 
 ## 运行
@@ -173,31 +173,6 @@ webui4frpc（单二进制，无 frp 依赖，前端 go:embed）
 
 集群页展示：环拓扑（self/leader 标记）、令牌轮次、节点负载、待办任务、活跃转发拓扑（含 owner 归属）、增量事件日志。
 
-## 集群演示
-
-```bash
-# 构建
-go build -o deploy/artifacts/webui4frpc ./cmd/webui4frpc
-
-# 拉起 4 节点集群 + 1 孤立节点（Docker）
-cd deploy && bash run-cluster.sh
-
-# 运行转发测试（提交 4 条转发 → 验证分发 → 撤销）
-bash test-forward.sh
-```
-
-节点端口映射：
-
-| 节点 | 端口 | 角色 |
-|---|---|---|
-| node-a | 7501 | 集群 leader（seed，有 frpc 缓存） |
-| node-b | 7502 | 集群成员 |
-| node-c | 7503 | 集群成员 |
-| node-d | 7504 | 集群成员 |
-| node-e | 7505 | 孤立节点（standalone，未加入集群） |
-
-所有节点 `-user=admin -password=admin123`。
-
 ## 测试
 
 ```bash

deploy/Dockerfile

@@ -1,21 +0,0 @@
-# Build the Go binary with embedded web UI + demo backend.
-FROM golang:1.25.0 AS build
-WORKDIR /src
-COPY go.mod go.sum vendor ./
-COPY cmd ./cmd
-COPY internal ./internal
-# Vendored deps => fully offline build.
-RUN CGO_ENABLED=0 go build -mod=vendor -trimpath -ldflags="-s -w" -o /out/webui4frpc ./cmd/webui4frpc
-
-# Static demo backend (tiny HTTP server).
-COPY deploy/demo-backend ./demo-backend
-RUN CGO_ENABLED=0 go build -mod=vendor -trimpath -ldflags="-s -w" -o /out/backend ./demo-backend
-
-# Runtime image: no frp binaries baked in (mounted via compose volume to keep
-# layers small and build fast); static Go only.
-FROM debian:bookworm-slim
-WORKDIR /app
-COPY --from=build /out/webui4frpc /app/webui4frpc
-COPY --from=build /out/backend /app/backend
-EXPOSE 7500 7000 7400 8080
-ENTRYPOINT ["/app/webui4frpc"]

deploy/README.md

@@ -1,46 +0,0 @@
-# webui4frpc Docker 集群演示
-
-一键起一套 **frps + backend + 3×webui4frpc 节点**（node-a 种子，node-b/c 通过 `-peer` 加入并自动从邻居拉取 frpc 二进制）。
-
-## 前置
-
-- Docker 可用（无需 build：用宿主编译产物挂载进容器）
-- 宿主机有 `/home/jianf/Program/frp/bin/{frpc,frps}`（或设置 `FRP=` 指向别的目录）
-
-## 启动/停止
-
-```bash
-cd deploy
-go build -o artifacts/webui4frpc ../cmd/webui4frpc   # 首次/改代码后重建
-go build -o artifacts/backend ./demo-backend
-bash run-cluster.sh          # 启动集群
-bash run-cluster.sh stop     # 停止
-```
-
-## 拓扑
-
-```
-               ┌──────────┐    ┌───────────────────────┐
-               │ w4f-frps │◄───│ node-a (7501, 种子)    │
-  tunnel:      │ 7000/7080/18081 ├── node-b (7502, peer a)│
-  http://127.0.0.1:18081 └──────────┘    └── node-c (7503, peer b)│
-      ▼                     └───────────────────────┘
-  w4f-backend (8080)  ← frpc 转发
-```
-
-## 演示点
-
-- **M6 二进制分发**：node-a 预置 `bin/frpc-0.70.1/frpc`；node-b/c 无 frpc，入网后自动向邻居拉取并落盘缓存
-- **M3 状态**：node-a 配置 `adminPort`，状态页显示真实 per-proxy 状态（running）
-- **隧道**：node-a 画布 local=backend:8080 × remote=frps1，frpc worker 启动后经 frps:18081 访问 backend
-
-## 验证
-
-```bash
-# 二进制分发
-curl -s -u admin:admin123 http://127.0.0.1:7502/api/manager/cluster/nodes
-# 隧道
-curl http://127.0.0.1:18081/   # -> hello from w4f demo backend
-# 代理状态
-curl -s -u admin:admin123 http://127.0.0.1:17401/api/status
-```

deploy/compose.yaml

@@ -1,114 +0,0 @@
-services:
-  frps:
-    image: w4f-demo:latest
-    container_name: w4f-frps
-    command: ["/app/frp/frps", "-c", "/app/frps.toml"]
-    volumes:
-      - ./frps.toml:/app/frps.toml:ro
-      - /home/jianf/Program/frp/bin:/app/frp:ro
-    ports:
-      - "7000:7000" # frps control
-      - "7080:7080" # vhostHTTPPort
-    networks:
-      - w4f
-
-  frps2:
-    image: w4f-demo:latest
-    container_name: w4f-frps2
-    command: ["/app/frp/frps", "-c", "/app/frps.toml"]
-    volumes:
-      - ./frps2.toml:/app/frps.toml:ro
-      - /home/jianf/Program/frp/bin:/app/frp:ro
-    ports:
-      - "7001:7001" # frps control
-      - "7081:7081" # vhostHTTPPort
-    networks:
-      - w4f
-
-  frps3:
-    image: w4f-demo:latest
-    container_name: w4f-frps3
-    command: ["/app/frp/frps", "-c", "/app/frps.toml"]
-    volumes:
-      - ./frps3.toml:/app/frps.toml:ro
-      - /home/jianf/Program/frp/bin:/app/frp:ro
-    ports:
-      - "7002:7002" # frps control
-      - "7082:7082" # vhostHTTPPort
-    networks:
-      - w4f
-
-  backend:
-    image: w4f-demo:latest
-    container_name: w4f-backend
-    command: ["/app/backend"]
-    ports:
-      - "8080:8080"
-    networks:
-      - w4f
-
-  node-a:
-    image: w4f-demo:latest
-    container_name: w4f-node-a
-    command:
-      - "-addr=0.0.0.0:7500"
-      - "-user=admin"
-      - "-password=admin123"
-      - "-workdir=/data"
-      - "-frpc=/app/frp/frpc"
-    volumes:
-      - ./node-a:/data
-      - /home/jianf/Program/frp/bin:/app/frp:ro
-    ports:
-      - "7501:7500"
-      - "17401:7400"
-    depends_on:
-      - frps
-    networks:
-      - w4f
-
-  node-b:
-    image: w4f-demo:latest
-    container_name: w4f-node-b
-    command:
-      - "-addr=0.0.0.0:7500"
-      - "-user=admin"
-      - "-password=admin123"
-      - "-workdir=/data"
-      - "-frpc=/app/frp/frpc"
-      - "-peer=node-a:7500"
-    volumes:
-      - ./node-b:/data
-      - /home/jianf/Program/frp/bin:/app/frp:ro
-    ports:
-      - "7502:7500"
-    depends_on:
-      - frps
-      - node-a
-    networks:
-      - w4f
-
-  node-c:
-    image: w4f-demo:latest
-    container_name: w4f-node-c
-    command:
-      - "-addr=0.0.0.0:7500"
-      - "-user=admin"
-      - "-password=admin123"
-      - "-workdir=/data"
-      - "-frpc=/app/frp/frpc"
-      - "-peer=node-b:7500"
-    volumes:
-      - ./node-c:/data
-      - /home/jianf/Program/frp/bin:/app/frp:ro
-    ports:
-      - "7503:7500"
-    depends_on:
-      - frps
-      - node-b
-    networks:
-      - w4f
-
-networks:
-  w4f:
-    driver: bridge

deploy/demo-backend/main.go

@@ -1,16 +0,0 @@
-package main
-
-import (
-	"fmt"
-	"net/http"
-)
-
-func main() {
-	http.HandleFunc("/healthz", func(w http.ResponseWriter, r *http.Request) {
-		w.WriteHeader(http.StatusOK)
-	})
-	http.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
-		fmt.Fprintf(w, "hello from w4f demo backend (path=%s)", r.URL.Path)
-	})
-	_ = http.ListenAndServe("0.0.0.0:8080", nil)
-}

deploy/frps.toml

@@ -1,16 +0,0 @@
-# frps demo config for the cluster showcase.
-bindAddr = "0.0.0.0"
-bindPort = 7000
-
-auth.method = "token"
-auth.token = "demo-token"
-
-# HTTP virtual host port (for http proxies)
-vhostHTTPPort = 7080
-
-webServer.addr = "0.0.0.0"
-webServer.port = 7500
-
-log.to = "/app/frps.log"
-log.level = "info"
-log.maxDays = 3

deploy/frps2.toml

@@ -1,16 +0,0 @@
-# Mock frps #2 for forwarding dispatch/enablement testing.
-# Same auth token as the primary frps so a single frpc worker config works
-# against any of them; distinct bindPort / vhostHTTPPort so multiple can run
-# on the same docker network without clashing.
-bindAddr = "0.0.0.0"
-bindPort = 7001
-
-auth.method = "token"
-auth.token = "demo-token"
-
-# HTTP virtual host port (for http proxies)
-vhostHTTPPort = 7081
-
-log.to = "/app/frps.log"
-log.level = "info"
-log.maxDays = 3

deploy/frps3.toml

@@ -1,16 +0,0 @@
-# Mock frps #3 for forwarding dispatch/enablement testing.
-# Same auth token as the primary frps so a single frpc worker config works
-# against any of them; distinct bindPort / vhostHTTPPort so multiple can run
-# on the same docker network without clashing.
-bindAddr = "0.0.0.0"
-bindPort = 7002
-
-auth.method = "token"
-auth.token = "demo-token"
-
-# HTTP virtual host port (for http proxies)
-vhostHTTPPort = 7082
-
-log.to = "/app/frps.log"
-log.level = "info"
-log.maxDays = 3

deploy/run-cluster.sh

@@ -1,60 +0,0 @@
-#!/usr/bin/env bash
-# w4f demo cluster: 4-node ring (a=leader, b/c/d join) + 1 isolated standalone (e).
-set -euo pipefail
-cd "$(dirname "$0")"
-
-FRP=${FRP:-/home/jianf/Program/frp/bin}
-NET=w4f-demo
-VER=0.70.1
-
-docker network inspect $NET >/dev/null 2>&1 || docker network create $NET
-
-# clean + seed node-a cache with real frpc binary
-for d in node-a node-b node-c node-d node-e; do rm -rf data/$d 2>/dev/null || true; mkdir -p data/$d; done
-mkdir -p data/node-a/bin/frpc-$VER
-[ -x data/node-a/bin/frpc-$VER/frpc ] || cp $FRP/frpc data/node-a/bin/frpc-$VER/frpc
-chmod +x data/node-a/bin/frpc-$VER/frpc
-
-echo "== starting backend =="
-docker run -d --name w4f-backend --user $(id -u):$(id -g) --network $NET --network-alias backend -p 18080:8080 -v $PWD/artifacts/backend:/app/backend:ro --restart unless-stopped --entrypoint /app/backend debian:bookworm-slim
-
-echo "== starting frps =="
-docker run -d --name w4f-frps --user $(id -u):$(id -g) --network $NET --network-alias frps -p 7000:7000 -p 7080:7080 -p 18081:18081 -v $FRP/frps:/app/frps:ro -v $PWD/frps.toml:/app/frps.toml:ro --restart unless-stopped --entrypoint /app/frps debian:bookworm-slim -c /app/frps.toml
-
-echo "== starting mock frps2 (bindPort 7001) =="
-docker run -d --name w4f-frps2 --user $(id -u):$(id -g) --network $NET --network-alias frps2 -p 7001:7001 -p 7081:7081 -v $FRP/frps:/app/frps:ro -v $PWD/frps2.toml:/app/frps.toml:ro --restart unless-stopped --entrypoint /app/frps debian:bookworm-slim -c /app/frps.toml
-
-echo "== starting mock frps3 (bindPort 7002) =="
-docker run -d --name w4f-frps3 --user $(id -u):$(id -g) --network $NET --network-alias frps3 -p 7002:7002 -p 7082:7082 -v $FRP/frps:/app/frps:ro -v $PWD/frps3.toml:/app/frps.toml:ro --restart unless-stopped --entrypoint /app/frps debian:bookworm-slim -c /app/frps.toml
-
-echo "== starting node-a (seed, has frpc cache) =="
-docker run -d --name w4f-node-a --user $(id -u):$(id -g) --network $NET --network-alias node-a -e W4F_HOST=node-a -p 7501:7500 -p 17401:7400 -v $PWD/artifacts/webui4frpc:/app/webui4frpc:ro -v $PWD/data/node-a:/data -v $FRP:/app/frp:ro --restart unless-stopped --entrypoint /app/webui4frpc debian:bookworm-slim -addr=0.0.0.0:7500 -user=admin -password=admin123 -workdir=/data -frpc=/app/frp/frpc
-
-# Wait for node-a to be ready and fetch its admission key. Nodes b/c/d join
-# via node-a, so they need node-a's nodeKey (-join-key) to pass the sponsor
-# verification in handleClusterJoin (403 on mismatch/empty). Node-e is isolated.
-echo "== fetching node-a join key =="
-KEY=""
-for _ in $(seq 1 15); do
-  KEY=$(curl -fsS -u admin:admin123 http://localhost:7501/api/manager/cluster/ring 2>/dev/null \
-    | python3 -c "import sys,json; print(json.load(sys.stdin).get('nodeKey',''))" 2>/dev/null || true)
-  [ -n "$KEY" ] && break
-  sleep 1
-done
-if [ -z "$KEY" ]; then echo "FATAL: could not fetch node-a join key"; docker logs w4f-node-a 2>&1 | tail 20; exit 1; fi
-echo "  node-a join key: $KEY"
-
-echo "== starting node-b (no frpc; joins via node-a) =="
-docker run -d --name w4f-node-b --user $(id -u):$(id -g) --network $NET --network-alias node-b -e W4F_HOST=node-b -p 7502:7500 -v $PWD/artifacts/webui4frpc:/app/webui4frpc:ro -v $PWD/data/node-b:/data -v $FRP:/app/frp:ro --restart unless-stopped --entrypoint /app/webui4frpc debian:bookworm-slim -addr=0.0.0.0:7500 -user=admin -password=admin123 -workdir=/data -frpc=/app/frp/frpc -peer=node-a:7500 -join-key=$KEY
-
-echo "== starting node-c (no frpc; joins via node-a) =="
-docker run -d --name w4f-node-c --user $(id -u):$(id -g) --network $NET --network-alias node-c -e W4F_HOST=node-c -p 7503:7500 -v $PWD/artifacts/webui4frpc:/app/webui4frpc:ro -v $PWD/data/node-c:/data -v $FRP:/app/frp:ro --restart unless-stopped --entrypoint /app/webui4frpc debian:bookworm-slim -addr=0.0.0.0:7500 -user=admin -password=admin123 -workdir=/data -frpc=/app/frp/frpc -peer=node-a:7500 -join-key=$KEY
-
-echo "== starting node-d (no frpc; joins via node-a) =="
-docker run -d --name w4f-node-d --user $(id -u):$(id -g) --network $NET --network-alias node-d -e W4F_HOST=node-d -p 7504:7500 -v $PWD/artifacts/webui4frpc:/app/webui4frpc:ro -v $PWD/data/node-d:/data -v $FRP:/app/frp:ro --restart unless-stopped --entrypoint /app/webui4frpc debian:bookworm-slim -addr=0.0.0.0:7500 -user=admin -password=admin123 -workdir=/data -frpc=/app/frp/frpc -peer=node-a:7500 -join-key=$KEY
-
-echo "== starting node-e (isolated standalone, not in cluster) =="
-docker run -d --name w4f-node-e --user $(id -u):$(id -g) --network $NET --network-alias node-e -e W4F_HOST=node-e -p 7505:7500 -v $PWD/artifacts/webui4frpc:/app/webui4frpc:ro -v $PWD/data/node-e:/data -v $FRP:/app/frp:ro --restart unless-stopped --entrypoint /app/webui4frpc debian:bookworm-slim -addr=0.0.0.0:7500 -user=admin -password=admin123 -workdir=/data -frpc=/app/frp/frpc
-
-echo "== cluster up =="
-docker ps --filter "name=w4f-" --format "table {{.Names}}\t{{.Status}}\t{{.Ports}}"

deploy/start-mock-frps.sh

@@ -1,33 +0,0 @@
-#!/usr/bin/env bash
-# Start (or restart) just the mock frps2/frps3 containers on the demo network,
-# leaving the rest of the cluster untouched. Used by test-forward.sh and when
-# the cluster is already running but the mock frps nodes are not.
-set -euo pipefail
-cd "$(dirname "$0")"
-
-FRP=${FRP:-/home/jianf/Program/frp/bin}
-NET=${W4F_NET:-w4f-demo}
-
-# detect which network the existing nodes live on (compose uses "w4f")
-if docker inspect w4f-node-a >/dev/null 2>&1; then
-  NET=$(docker inspect -f '{{range $k,$v := .NetworkSettings.Networks}}{{$k}}{{end}}' w4f-node-a 2>/dev/null || echo "$NET")
-fi
-
-start_one() {
-  local name="$1" alias="$2" cfg="$3" bp="$4" vp="$5"
-  docker rm -f "$name" >/dev/null 2>&1 || true
-  docker run -d --name "$name" --user "$(id -u):$(id -g)" \
-    --network "$NET" --network-alias "$alias" \
-    -p "$bp:$bp" -p "$vp:$vp" \
-    -v "$FRP/frps:/app/frps:ro" -v "$PWD/$cfg:/app/frps.toml:ro" \
-    --restart unless-stopped --entrypoint /app/frps \
-    debian:bookworm-slim -c /app/frps.toml >/dev/null
-  echo "  $name up ($alias, bindPort $bp, vhost $vp) on network $NET"
-}
-
-echo "== starting mock frps nodes on network $NET =="
-start_one w4f-frps2 frps2 frps2.toml 7001 7081
-start_one w4f-frps3 frps3 frps3.toml 7002 7082
-
-echo "== mock frps nodes up =="
-docker ps --filter "name=w4f-frps" --format "table {{.Names}}\t{{.Status}}\t{{.Ports}}"

deploy/test-forward.sh

@@ -1,230 +0,0 @@
-#!/usr/bin/env bash
-# test-forward.sh — exercise the token-ring forwarding dispatch pipeline.
-#
-# Submits 4 forwards via PUT /canvas (localOnly=false → ring tasks):
-#   r-frps  → frps:7000   (real, expect connected)
-#   r-frps2 → frps2:7001  (real mock, expect connected)
-#   r-frps3 → frps3:7002  (real mock, expect connected)
-#   r-dead  → 192.0.2.1    (dead addr, expect dispatch+worker-start, conn failed)
-#
-# Verifies: pending→0, topology claimed (with ownerId), worker running on the
-# claiming node, frpc "login to server success" for real targets, dead target
-# worker running but conn failed. Then revokes (restore canvas) and verifies
-# topology cleared + worker stopped.
-#
-# Run after the cluster is up: bash deploy/test-forward.sh
-set -euo pipefail
-cd "$(dirname "$0")"
-
-AUTH=admin:admin123
-A=http://localhost:7501
-B=http://localhost:7502
-C=http://localhost:7503
-D=http://localhost:7504
-E=http://localhost:7505
-NODE_PORTS=("$A" "$B" "$C" "$D" "$E")
-RESULTS=$(mktemp -t w4f-results.XXXX)
-trap 'rm -f "$RESULTS"' EXIT
-
-# ---- helpers ----
-cget() { curl -s -u "$AUTH" "$1/api/manager$2"; }
-cput() { curl -s -X PUT -u "$AUTH" -H 'Content-Type: application/json' --data "$3" "$1/api/manager$2"; }
-cpost() { curl -s -X POST -u "$AUTH" -H 'Content-Type: application/json' --data "$3" "$1/api/manager$2"; }
-# py '<stmt>' reads stdin json as d; stmt may use d
-py() { python3 -c "import sys,json; d=json.load(sys.stdin); print($1)"; }
-
-ok()   { printf "  \033[32m✓\033[0m %s\n" "$1"; }
-fail() { printf "  \033[31m✗\033[0m %s\n" "$1"; }
-info() { printf "  · %s\n" "$1"; }
-
-echo "╭──────────────────────────────────────────────╮"
-echo "│  w4f forwarding dispatch pipeline test        │"
-echo "╰──────────────────────────────────────────────╯"
-
-# ---- preflight: cluster + mock frps up ----
-if ! docker ps --format '{{.Names}}' | grep -q '^w4f-node-a$'; then
-  fail "w4f-node-a not running — start the cluster first: bash deploy/run-cluster.sh"
-  exit 1
-fi
-bash "$PWD/start-mock-frps.sh" >/dev/null 2>&1 || true
-
-# ---- 1. confirm ring has 5 nodes (wait for convergence) ----
-echo "[1] cluster ring health (expect 4 nodes)"
-deadline=$(( $(date +%s) + 30 ))
-NN=0
-while [ $(date +%s) -lt $deadline ]; do
-  RING=$(cget "$A" /cluster/ring)
-  NN=$(py 'len(d.get("nodes",[]))' <<< "$RING")
-  if [ "$NN" -ge 4 ]; then break; fi
-  sleep 1
-done
-if [ "$NN" -ge 4 ]; then ok "ring has $NN nodes"; else fail "ring has $NN nodes (expected ≥4)"; exit 1; fi
-
-# ---- 1.5 join key security (wrong key → 403, correct key present) ----
-echo "[1.5] join key security"
-KEY=$(py 'd.get("nodeKey","")' <<< "$RING")
-if [ -n "$KEY" ]; then ok "nodeKey present in snapshot (${KEY:0:8}…)"; else fail "no nodeKey in ring snapshot"; fi
-CODE=$(curl -s -o /dev/null -w "%{http_code}" -X POST -u "$AUTH" -H 'Content-Type: application/json' \
-  --data '{"id":"evil","addr":"evil:7500","version":"0.1.0","joinKey":"wrong-key"}' \
-  "$A/api/manager/cluster/join")
-if [ "$CODE" = "403" ]; then ok "wrong join key → 403 (rejected)"; else fail "wrong join key → HTTP $CODE (expected 403)"; fi
-# empty key → 403 too
-CODE2=$(curl -s -o /dev/null -w "%{http_code}" -X POST -u "$AUTH" -H 'Content-Type: application/json' \
-  --data '{"id":"evil","addr":"evil:7500","version":"0.1.0","joinKey":""}' \
-  "$A/api/manager/cluster/join")
-if [ "$CODE2" = "403" ]; then ok "empty join key → 403 (rejected)"; else fail "empty join key → HTTP $CODE2 (expected 403)"; fi
-
-# ---- 2. snapshot original canvas (for cleanup/restore) ----
-echo "[2] snapshot canvas"
-ORIG=$(cget "$A" /canvas)
-CNT=$(echo "$ORIG" | py 'len((d.get("locals") or [])) + len((d.get("remotes") or [])) + len((d.get("links") or []))')
-info "original canvas: $CNT entries (locals+remotes+links)"
-
-# ---- 3. build merged canvas with 4 forwards + PUT ----
-echo "[3] submit 4 forwards (PUT /canvas on node-a)"
-MERGED=$(python3 -c '
-import json, sys
-orig = json.load(sys.stdin)
-locals_ = orig.get("locals", []) or []
-remotes = orig.get("remotes", []) or []
-links   = orig.get("links", []) or []
-
-# de-dup by name so re-runs are idempotent
-have_l = {l["name"] for l in locals_}
-have_r = {r["name"] for r in remotes}
-if "backend-svc" not in have_l:
-    locals_.append({"name":"backend-svc","ip":"backend","port":8080,"protocol":"tcp","localOnly":False})
-
-new_rems = [
-    ("r-frps",  "frps",      7000),
-    ("r-frps2", "frps2",     7001),
-    ("r-frps3", "frps3",     7002),
-    ("r-dead",  "192.0.2.1", 7000),
-]
-for nm, ip, port in new_rems:
-    if nm not in have_r:
-        remotes.append({"name":nm,"ip":ip,"port":port,"token":"demo-token","enabled":True})
-
-ports = {"r-frps":18091,"r-frps2":18092,"r-frps3":18093,"r-dead":18094}
-have_link = {(l["local"],l["remote"]) for l in links}
-for nm in ports:
-    key = ("backend-svc", nm)
-    if key not in have_link:
-        links.append({"local":"backend-svc","remote":nm,"remotePort":ports[nm]})
-
-print(json.dumps({"locals":locals_,"remotes":remotes,"links":links}))
-' <<< "$ORIG")
-
-cput "$A" /canvas "$MERGED" >/dev/null
-ok "canvas submitted (1 local → 4 remotes, localOnly=false)"
-
-# the 4 remotes we expect to see claimed
-declare -A EXPECT_PORT=( [r-frps]=18091 [r-frps2]=18092 [r-frps3]=18093 [r-dead]=18094 )
-declare -A EXPECT_CONN=( [r-frps]=connected [r-frps2]=connected [r-frps3]=connected [r-dead]=failed )
-declare -A OWNER=      # remote name -> node port URL
-
-# ---- 4. poll ring: pending → 0 + topology has our 4 ----
-echo "[4] wait for dispatch: pending → claimed into topology"
-deadline=$(( $(date +%s) + 45 ))
-claimed=0
-while [ $(date +%s) -lt $deadline ]; do
-  RING=$(cget "$A" /cluster/ring)
-  pend=$(py 'len(d.get("pending",[]))' <<< "$RING")
-  # count our remotes present in topology (python — robust vs empty arrays)
-  cnt=$(py 'sum(1 for t in d.get("topology",[]) if t["remote"]["name"] in ("r-frps","r-frps2","r-frps3","r-dead"))' <<< "$RING")
-  if [ "$cnt" -ge 4 ]; then ok "all 4 in topology (pending=$pend)"; claimed=1; break; fi
-  sleep 1
-done
-if [ "$claimed" -ne 1 ]; then
-  fail "only $cnt/4 reached topology after 45s (pending=$pend)"
-  RING=$(cget "$A" /cluster/ring); info "ring: $(py 'json.dumps({"pending":len(d.get("pending",[])),"topo":[t["remote"]["name"] for t in d.get("topology",[])]})' <<< "$RING")"
-  exit 1
-fi
-info "topology: $(cget "$A" /cluster/ring | py '",".join(t["remote"]["name"]+"@"+t["ownerId"][:8] for t in d.get("topology",[]))')"
-
-# ---- 5. find owner node per remote + assert worker running ----
-echo "[5] locate claiming node + assert worker running"
-for nm in r-frps r-frps2 r-frps3 r-dead; do
-  found=""
-  deadline=$(( $(date +%s) + 40 ))
-  while [ $(date +%s) -lt $deadline ]; do
-    for P in "${NODE_PORTS[@]}"; do
-      ST=$(cget "$P" /status)
-      # does this node have a profile named $nm with process.state=running?
-      run=$(py 'next((1 for p in d.get("profiles",[]) if p["name"]=="'"$nm"'" and p.get("process",{}).get("state")=="running"),0)' <<< "$ST")
-      if [ "$run" = "1" ]; then found="$P"; break; fi
-    done
-    [ -n "$found" ] && break
-    sleep 1
-  done
-  if [ -n "$found" ]; then
-    OWNER[$nm]="$found"
-    ok "$nm → worker running on $(echo "$found" | sed 's#http://localhost:##')"
-  else
-    fail "$nm → no running worker found on any node within 40s"
-    OWNER[$nm]=""
-  fi
-done
-
-# ---- 6. connection state + frpc login log ----
-echo "[6] connection state + frpc login log"
-for nm in r-frps r-frps2 r-frps3 r-dead; do
-  P="${OWNER[$nm]}"
-  [ -z "$P" ] && { fail "$nm skipped (no worker)"; continue; }
-  ST=$(cget "$P" /status)
-  conn=$(py 'next((p.get("connState","") for p in d.get("profiles",[]) if p["name"]=="'"$nm"'"),"")' <<< "$ST")
-  expect="${EXPECT_CONN[$nm]}"
-  if [ "$nm" = "r-dead" ]; then
-    # dead: worker running, conn NOT connected
-    if [ "$conn" != "connected" ]; then ok "$nm connState=$conn (dead, not connected — correct)"; else fail "$nm unexpectedly connected"; fi
-  else
-    # real: wait for connected + login log
-    deadline=$(( $(date +%s) + 40 ))
-    logged=""
-    while [ $(date +%s) -lt $deadline ]; do
-      LOGS=$(cget "$P" /profiles/$nm/logs)
-      if echo "$LOGS" | grep -q "login to server success"; then logged=1; break; fi
-      sleep 1
-    done
-    if [ -n "$logged" ]; then ok "$nm frpc logged in (login to server success)"; else fail "$nm no login-success in 40s (connState=$conn)"; fi
-  fi
-done
-
-# ---- 6.5 tunnel data path for real frps (bonus) ----
-echo "[6.5] tunnel data path (real frps → backend)"
-for nm in r-frps r-frps2 r-frps3; do
-  port="${EXPECT_PORT[$nm]}"
-  # curl the remotePort from inside the node-a container (same docker net)
-  body=$(docker exec w4f-node-a sh -c "command -v curl >/dev/null 2>&1 && curl -sS --max-time 4 http://frps$([ "$nm" != "r-frps" ] && echo "2" || echo ""):$port 2>/dev/null || wget -qO- --timeout=4 http://frps$([ "$nm" != "r-frps" ] && echo "2" || echo ""):$port 2>/dev/null" 2>/dev/null || true)
-  # the backend responds with some body; just check non-empty or a hit
-  if [ -n "$body" ]; then ok "$nm tunnel :$port → backend returned data"; else info "$nm tunnel :$port no body (may be expected if backend gives empty)"; fi
-done
-
-# ---- 7. revoke (restore original canvas) ----
-echo "[7] revoke: restore original canvas → wait for topology cleared"
-cput "$A" /canvas "$ORIG" >/dev/null
-ok "canvas restored to original snapshot"
-deadline=$(( $(date +%s) + 30 ))
-cleared=0
-while [ $(date +%s) -lt $deadline ]; do
-  RING=$(cget "$A" /cluster/ring)
-  cnt=$(py 'sum(1 for t in d.get("topology",[]) if t["remote"]["name"] in ("r-frps","r-frps2","r-frps3","r-dead"))' <<< "$RING")
-  if [ "$cnt" -eq 0 ]; then ok "topology cleared of all 4 forwards"; cleared=1; break; fi
-  sleep 1
-done
-[ "$cleared" -ne 1 ] && fail "topology still has $cnt forwards after 30s"
-
-# assert workers stopped on owners
-for nm in r-frps r-frps2 r-frps3 r-dead; do
-  P="${OWNER[$nm]}"; [ -z "$P" ] && continue
-  ST=$(cget "$P" /status)
-  run=$(py 'next((1 for p in d.get("profiles",[]) if p["name"]=="'"$nm"'" and p.get("process",{}).get("state")=="running"),0)' <<< "$ST")
-  if [ "$run" = "0" ]; then ok "$nm worker stopped on $(echo "$P" | sed 's#http://localhost:##')"; else fail "$nm worker still running after revoke"; fi
-done
-
-# ---- summary ----
-echo ""
-echo "╭── result ──╮"
-echo "│ dispatch + claim + worker + connected : verified (3 real frps + 1 dead) "
-echo "│ revoke (canvas restore) → topology cleared + workers stopped : verified   "
-echo "╰──────────────────────────────────────────────────────────────────────────╯"