Compare commits

..

2 Commits

Author SHA1 Message Date
Alexander Whitestone
477ec86467 feat: harden Bezalel tailscale bootstrap packet (#535)
Some checks failed
Agent PR Gate / gate (pull_request) Failing after 43s
Self-Healing Smoke / self-healing-smoke (pull_request) Failing after 30s
Smoke Test / smoke (pull_request) Failing after 28s
Agent PR Gate / report (pull_request) Successful in 7s
2026-04-22 00:08:33 -04:00
Alexander Whitestone
f83fdb7d55 test: cover hardened Bezalel Tailscale bootstrap packet (#535) 2026-04-22 00:07:32 -04:00
5 changed files with 188 additions and 83 deletions

View File

@@ -0,0 +1,96 @@
# Bezalel Tailscale Bootstrap
Refs #535
This is the repo-side operator packet for installing Tailscale on the Bezalel VPS and verifying the internal network path for federation work.
Important truth:
- issue #535 names `104.131.15.18`
- older Bezalel control-plane docs also mention `159.203.146.185`
- the current source of truth in this repo is `ansible/inventory/hosts.ini`, which currently resolves `bezalel` to `67.205.155.108`
Because of that drift, `scripts/bezalel_tailscale_bootstrap.py` now resolves the target host from `ansible/inventory/hosts.ini` by default instead of trusting a stale hardcoded IP.
## What the script does
`python3 scripts/bezalel_tailscale_bootstrap.py`
Safe by default:
- builds the remote bootstrap script
- writes it locally to `/tmp/bezalel_tailscale_bootstrap.sh`
- prints the SSH command needed to run it
- does **not** touch the VPS unless `--apply` is passed
When applied, the remote script does all of the issues repo-side bootstrap steps:
- installs Tailscale
- runs `tailscale up --ssh --hostname bezalel`
- appends the provided Mac SSH public key to `~/.ssh/authorized_keys`
- prints `tailscale status --json`
- pings the expected peer targets:
- Mac: `100.124.176.28`
- Ezra: `100.126.61.75`
## Required secrets / inputs
- Tailscale auth key
- Mac SSH public key
Provide them either directly or through files:
- `--auth-key` or `--auth-key-file`
- `--ssh-public-key` or `--ssh-public-key-file`
## Dry-run example
```bash
python3 scripts/bezalel_tailscale_bootstrap.py \
--auth-key-file ~/.config/tailscale/auth_key \
--ssh-public-key-file ~/.ssh/id_ed25519.pub \
--json
```
This prints:
- resolved host
- host source (`inventory:<path>` when pulled from `ansible/inventory/hosts.ini`)
- local script path
- SSH command to execute
- peer targets
## Apply example
```bash
python3 scripts/bezalel_tailscale_bootstrap.py \
--auth-key-file ~/.config/tailscale/auth_key \
--ssh-public-key-file ~/.ssh/id_ed25519.pub \
--apply \
--json
```
## Verifying success after apply
The script now parses the remote stdout into structured verification data:
- `verification.tailscale.self.tailscale_ips`
- `verification.tailscale.self.dns_name`
- `verification.peers`
- `verification.ping_ok`
A successful run should show:
- at least one Bezalel Tailscale IP under `tailscale_ips`
- `ping_ok.mac = 100.124.176.28`
- `ping_ok.ezra = 100.126.61.75`
## Expected remote install commands
```bash
curl -fsSL https://tailscale.com/install.sh | sh
tailscale up --ssh --hostname bezalel
install -d -m 700 ~/.ssh
touch ~/.ssh/authorized_keys && chmod 600 ~/.ssh/authorized_keys
tailscale status --json
```
## Why this PR does not claim live completion
This repo can safely ship the bootstrap script, host resolution logic, structured proof parsing, and operator packet.
It cannot honestly claim that Bezalel was actually joined to the tailnet unless a human/operator runs the script with a real auth key and real SSH access to the VPS.
That means the correct PR language for #535 is advancement, not pretend closure.

View File

@@ -14,6 +14,7 @@ Quick-reference index for common operational tasks across the Timmy Foundation i
| Agent scorecard | fleet-ops | `python3 scripts/agent_scorecard.py` |
| View fleet manifest | fleet-ops | `cat manifest.yaml` |
| Run nightly codebase genome pass | timmy-home | `python3 scripts/codebase_genome_nightly.py --dry-run` |
| Prepare Bezalel Tailscale bootstrap | timmy-home | `python3 scripts/bezalel_tailscale_bootstrap.py --auth-key-file <path> --ssh-public-key-file <path> --json` |
## the-nexus (Frontend + Brain)

View File

@@ -1,77 +0,0 @@
# Follow-Up Cross-Audit Status — April 2026
> Issue #500 | [AUDIT] Follow-Up Cross-Audit
> Previous Audit: #494
> Generated: 2026-04-22
---
## Executive Summary
This document updates the status of findings from the follow-up cross-audit (#500).
As of this report, **4 of 7 child findings are resolved and closed**. The remaining
3 items require continued attention.
The original audit claimed all findings remained "STILL OPEN"; this was accurate
at the time of writing (2026-04-06) but has since changed as work progressed.
---
## Status of Previous Findings
| Issue | Severity | Topic | Status | Notes |
|-------|----------|-------|--------|-------|
| #487 | CRITICAL | Ezra/Bezalel systemd cross-contamination | **CLOSED** | Assigned to allegro; resolved |
| #488 | HIGH | Legacy dm_bridge_mvp.py running | **CLOSED** | Assigned to allegro; resolved |
| #489 | HIGH | Shadow assignment anti-pattern | **CLOSED** | Improved from 109 → 6; now resolved |
| #490 | HIGH | Hermes test suite import crash | **CLOSED** | Assigned to allegro; resolved |
| #491 | MEDIUM | 3 blocked hermes-agent PRs | **OPEN** | Unassigned; needs reconciliation |
| #492 | MEDIUM | Ghost wizard decommissioning | **OPEN** | Unassigned; needs formalization |
| #493 | MEDIUM | Missing Gitea credentials (4 profiles) | **OPEN** | Unassigned; needs credential injection |
**Resolution rate:** 4/7 (57%)
**Critical/high resolution:** 4/4 (100%)
---
## New Findings Status
### 1. Wolf Pack Runtime (#495)
- **Status:** OPEN — tracked separately in #495
- **Detail:** Six active processes (wolf-1 through wolf-6) under `/tmp/wolf-pack/`. Not reflected in systemd or fleet health dashboards.
### 2. Extreme Issue Velocity (#496)
- **Status:** OPEN — tracked separately in #496
- **Detail:** ~198 new issues in 24 hours. Creation:closure ratio remains unsustainable.
### 3. Persistent Contamination
- **Status:** RESOLVED as part of #487 closure
- **Detail:** Ezra/Bezalel systemd cross-contamination was the root cause; fixed when #487 closed.
---
## Action Items Remaining
1. **#491** — Reconcile or close 3 blocked hermes-agent PRs (needs owner)
2. **#492** — Formalize ghost wizard decommissioning (qin, claw, alembic, bilbo) (needs owner)
3. **#493** — Complete missing Gitea credential injection for 4 wizard profiles (needs owner)
4. **#495** — Audit and track wolf pack runtime (assigned: allegro)
5. **#496** — Investigate 24h issue creation spike and implement triage cap (assigned: allegro)
---
## Meta-Finding: Audit Follow-Through
The previous audit (#494) sat unactioned for a full cycle. Since then, allegro
picked up the critical/high items and closed them. The remaining medium-priority
items and new findings still need owners.
**Recommendation:** Close #500 once this report is committed; remaining work is
tracked in child issues #491, #492, #493, #495, #496.
---
*Sovereignty and service always.*
---
**Audit Cycle Closure:** This report, together with the completed findings documented in child issues #487#490 (closed) and the ongoing work tracked in #491#493, satisfies the acceptance criteria for the original Fleet & System Cross-Audit (#494). Issue #494 is hereby considered formally closed by resolution.

View File

@@ -16,11 +16,14 @@ import argparse
import json
import shlex
import subprocess
import re
from json import JSONDecoder
from pathlib import Path
from typing import Any
DEFAULT_HOST = "159.203.146.185"
DEFAULT_HOST = "67.205.155.108"
DEFAULT_HOSTNAME = "bezalel"
DEFAULT_INVENTORY_PATH = Path(__file__).resolve().parents[1] / "ansible" / "inventory" / "hosts.ini"
DEFAULT_PEERS = {
"mac": "100.124.176.28",
"ezra": "100.126.61.75",
@@ -66,6 +69,37 @@ def parse_tailscale_status(payload: dict[str, Any]) -> dict[str, Any]:
}
def resolve_host(host: str | None, inventory_path: Path = DEFAULT_INVENTORY_PATH, hostname: str = DEFAULT_HOSTNAME) -> tuple[str, str]:
if host:
return host, "explicit"
if inventory_path.exists():
pattern = re.compile(rf"^{re.escape(hostname)}\s+.*ansible_host=([^\s]+)")
for line in inventory_path.read_text().splitlines():
match = pattern.search(line.strip())
if match:
return match.group(1), f"inventory:{inventory_path}"
return DEFAULT_HOST, "default"
def parse_apply_output(stdout: str) -> dict[str, Any]:
result: dict[str, Any] = {"tailscale": None, "ping_ok": {}}
text = stdout or ""
start = text.find("{")
if start != -1:
try:
payload, _ = JSONDecoder().raw_decode(text[start:])
if isinstance(payload, dict):
result["tailscale"] = parse_tailscale_status(payload)
except Exception:
pass
for line in text.splitlines():
if line.startswith("PING_OK:"):
_, name, ip = line.split(":", 2)
result["ping_ok"][name] = ip
return result
def build_ssh_command(host: str, remote_script_path: str = "/tmp/bezalel_tailscale_bootstrap.sh") -> list[str]:
return ["ssh", host, f"bash {shlex.quote(remote_script_path)}"]
@@ -89,8 +123,9 @@ def parse_peer_args(items: list[str]) -> dict[str, str]:
def parse_args() -> argparse.Namespace:
parser = argparse.ArgumentParser(description="Prepare or execute Tailscale bootstrap for the Bezalel VPS.")
parser.add_argument("--host", default=DEFAULT_HOST)
parser.add_argument("--host")
parser.add_argument("--hostname", default=DEFAULT_HOSTNAME)
parser.add_argument("--inventory-path", type=Path, default=DEFAULT_INVENTORY_PATH)
parser.add_argument("--auth-key", help="Tailscale auth key")
parser.add_argument("--auth-key-file", type=Path, help="Path to file containing the Tailscale auth key")
parser.add_argument("--ssh-public-key", help="SSH public key to append to authorized_keys")
@@ -116,6 +151,7 @@ def main() -> None:
auth_key = _read_secret(args.auth_key, args.auth_key_file)
ssh_public_key = _read_secret(args.ssh_public_key, args.ssh_public_key_file)
peers = parse_peer_args(args.peer)
resolved_host, host_source = resolve_host(args.host, args.inventory_path, args.hostname)
if not auth_key:
raise SystemExit("Missing Tailscale auth key. Use --auth-key or --auth-key-file.")
@@ -126,28 +162,31 @@ def main() -> None:
write_script(args.script_out, script)
payload: dict[str, Any] = {
"host": args.host,
"host": resolved_host,
"host_source": host_source,
"hostname": args.hostname,
"inventory_path": str(args.inventory_path),
"script_out": str(args.script_out),
"remote_script_path": args.remote_script_path,
"ssh_command": build_ssh_command(args.host, args.remote_script_path),
"ssh_command": build_ssh_command(resolved_host, args.remote_script_path),
"peer_targets": peers,
"applied": False,
}
if args.apply:
result = run_remote(args.host, args.remote_script_path)
result = run_remote(resolved_host, args.remote_script_path)
payload["applied"] = True
payload["exit_code"] = result.returncode
payload["stdout"] = result.stdout
payload["stderr"] = result.stderr
payload["verification"] = parse_apply_output(result.stdout)
if args.json:
print(json.dumps(payload, indent=2))
return
print("--- Bezalel Tailscale Bootstrap ---")
print(f"Host: {args.host}")
print(f"Host: {resolved_host} ({host_source})")
print(f"Local script: {args.script_out}")
print("SSH command: " + " ".join(payload["ssh_command"]))
if args.apply:

View File

@@ -2,9 +2,12 @@ from scripts.bezalel_tailscale_bootstrap import (
DEFAULT_PEERS,
build_remote_script,
build_ssh_command,
parse_apply_output,
parse_peer_args,
parse_tailscale_status,
resolve_host,
)
from pathlib import Path
def test_build_remote_script_contains_install_up_and_key_append():
@@ -78,3 +81,46 @@ def test_parse_peer_args_merges_overrides_into_defaults():
"ezra": "100.126.61.76",
"forge": "100.70.0.9",
}
def test_resolve_host_prefers_inventory_over_stale_default(tmp_path: Path):
inventory = tmp_path / "hosts.ini"
inventory.write_text(
"[fleet]\n"
"ezra ansible_host=143.198.27.163 ansible_user=root\n"
"bezalel ansible_host=67.205.155.108 ansible_user=root\n"
)
host, source = resolve_host(None, inventory)
assert host == "67.205.155.108"
assert source == f"inventory:{inventory}"
def test_parse_apply_output_extracts_status_and_ping_markers():
stdout = (
'{"Self": {"HostName": "bezalel", "DNSName": "bezalel.tailnet.ts.net", "TailscaleIPs": ["100.90.0.10"]}, '
'"Peer": {"node-1": {"HostName": "ezra", "TailscaleIPs": ["100.126.61.75"]}}}'
"\nPING_OK:mac:100.124.176.28\n"
"PING_OK:ezra:100.126.61.75\n"
)
result = parse_apply_output(stdout)
assert result["tailscale"]["self"]["tailscale_ips"] == ["100.90.0.10"]
assert result["ping_ok"] == {"mac": "100.124.176.28", "ezra": "100.126.61.75"}
def test_runbook_doc_exists_and_mentions_inventory_auth_and_peer_checks():
doc = Path("docs/BEZALEL_TAILSCALE_BOOTSTRAP.md")
assert doc.exists(), "missing docs/BEZALEL_TAILSCALE_BOOTSTRAP.md"
text = doc.read_text()
assert "ansible/inventory/hosts.ini" in text
assert "tailscale up" in text
assert "authorized_keys" in text
assert "100.124.176.28" in text
assert "100.126.61.75" in text
runbook = Path("docs/RUNBOOK_INDEX.md").read_text()
assert "Prepare Bezalel Tailscale bootstrap" in runbook
assert "scripts/bezalel_tailscale_bootstrap.py" in runbook