NetVM: fleet control script (netvm-fleet.sh) for operators

NODES.md

@@ -5,4 +5,4 @@ designed choice — record the reason.
 
 | node | netns | warp identity | egress IP | tail IP | shared with / reason |
 |------|-------|---------------|-----------|---------|----------------------|
-| laptop (node zero) | — | — | — | — | orchestrator + first node, not yet provisioned |
+| tp | warp-tp | — | — | — | laptop; orchestrator + first node; identity pending (netvm-new-identity.sh tp) |

README.md

@@ -83,6 +83,7 @@ Tear down: sudo bin/netvm-node-down.sh <node>.
 - bin/netvm-node-down.sh <node> — tear a node's egress down.
 - bin/netvm-topology.sh — print the live topology table.
 - `bin/netvm-provision-edge.sh` — prepare an edge device (sudoers allowlist, deps, /etc/netvm); run on the node, once.
+- `bin/netvm-fleet.sh` — operator fleet control over the tailnet (topology/up/down/ssh per node).
 - NODES.md — the registry: node -> netns -> Warp identity -> egress IP.
 
 ## Verification checklist

bin/netvm-fleet.sh

@@ -0,0 +1,59 @@
+#!/usr/bin/env bash
+# netvm-fleet.sh — operator fleet control over the tailnet.
+# Run on any tailnet-connected host (laptop, VM) as the operator user.
+#   netvm-fleet.sh topology        # egress per node, fleet-wide
+#   netvm-fleet.sh up <node>       # bring a node's egress up
+#   netvm-fleet.sh down <node>     # bring it down
+#   netvm-fleet.sh ssh <node>      # shell on the node
+# Node names are tailnet hostnames (see NODES.md).
+# Prereqs: targets provisioned via netvm-provision-edge.sh; this user's SSH
+# key accepted on targets. Lifecycle runs through the sudoers allowlist
+# (sudo -n), so it is audit-logged and never blanket root.
+set -euo pipefail
+REPO="${NETVM_REPO:-$HOME/Projects/NetVM}"
+OPERATOR_USER="${OPERATOR_USER:-$(whoami)}"
+
+nodes() {
+  awk -F'|' '/^\|/ && $2 !~ /node/ && $2 !~ /---/ { n=$2; gsub(/^ +| +$/, "", n); if (n != "") print n }' "$REPO/NODES.md"
+}
+
+tail_ip() {
+  NODE="$1" python3 -c "
+import json, os, subprocess, sys
+node = os.environ['NODE']
+st = json.loads(subprocess.run(['tailscale','status','--json'], capture_output=True, text=True).stdout)
+cands = list(st.get('Peer', {}).values()) + [st.get('Self', {})]
+for p in cands:
+    names = {p.get('HostName',''), p.get('DNSName','').split('.')[0]}
+    if node in names and p.get('TailscaleIPs'):
+        print(p['TailscaleIPs'][0]); sys.exit(0)
+sys.exit(1)
+"
+}
+
+run_on() { # <node> <remote-command...>
+  local node="$1"; shift
+  local ip
+  ip=$(tail_ip "$node") || { echo "unknown tailnet node: $node"; exit 1; }
+  ssh -o BatchMode=yes -o ConnectTimeout=15 "${OPERATOR_USER}@${ip}" "$@"
+}
+
+cmd="${1:?usage: netvm-fleet.sh topology|up <node>|down <node>|ssh <node>}"
+case "$cmd" in
+  topology)
+    for n in $(nodes); do
+      printf '== %s ==\n' "$n"
+      run_on "$n" 'sudo -n $HOME/Projects/NetVM/bin/netvm-topology.sh' 2>&1 || echo "unreachable"
+    done
+    ;;
+  up|down)
+    node="${2:?usage: netvm-fleet.sh $cmd <node>}"
+    run_on "$node" "sudo -n \$HOME/Projects/NetVM/bin/netvm-node-${cmd}.sh ${node}"
+    ;;
+  ssh)
+    node="${2:?usage: netvm-fleet.sh ssh <node>}"
+    ip=$(tail_ip "$node") || { echo "unknown tailnet node: $node"; exit 1; }
+    exec ssh "${OPERATOR_USER}@${ip}"
+    ;;
+  *) echo "unknown command: $cmd"; exit 1 ;;
+esac