feat: monky-deployd v0.1.0 — pull agent over the mesh (ADR-0028)

Stdlib-only Python 3.12 agent for docker VMs and laptops: flock → checkin
(bearer = the agent's OpenBao token, bootstrapped from the install kit's
jwt-tenancy deploy grant) → action apply|none|down → bundle (sha256
verified) → refusal checks (unresolved ${VAR} names only, manifest paths
pinned to monky/data/<env>/see/, privileged/host-network, rollback, disk
need×1.5+headroom) → lease → POST /v1/auth/jwt-tenancy/login → KV reads →
.env 0600 → promote → compose pull/up → wait healthy → report; finally
renew-self / re-lease before max TTL, scrub. Exit 0/75/78/1. Redactor log
filter. Transports sdk (openziti) / proxy (ziti tunnel proxy 18443/18200) /
system. Laptop mode.

Packaging: hardened oneshot + 60 s timer + proxy unit, nfpm .deb with
/opt/monky-deployd/venv, install.sh for Ubuntu 26.04 (Gitea release
download, enrol, ACLs, bootstrap from stdin), ansible role skeleton for
osg1-07. CI: lint/test on every change; wheel (openziti on ubuntu:26.04) and
package (nfpm) allow_failure until runner egress is proven; GitLab release +
release:gitea on v* tags. Docs: README, PROTOCOL, OPERATIONS, CHANGELOG,
CLAUDE/AGENTS.

Divergence noted: monky-tenancy main (MR !15) still ships the AppRole lease
and kit; this agent implements the plan's Gate 1 RESULT (login_jwt, no
unwrap) and refuses an AppRole lease loudly (LEASE_SHAPE).

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KLB7jieMNRkTsJ2epr4Ds1
This commit is contained in:
2026-09-05 08:01:36 +00:00
parent c72d6c0227
commit 1c42e913a8
48 changed files with 4932 additions and 64 deletions
View File
+113
View File
@@ -0,0 +1,113 @@
from __future__ import annotations
import json
import os
import sys
from pathlib import Path
import pytest
ROOT = Path(__file__).resolve().parents[1]
sys.path.insert(0, str(ROOT))
from monky_deployd import config as configmod # noqa: E402
from monky_deployd.redact import REDACTOR # noqa: E402
from tests.fakes import ENV, FakeBao, FakeTenancy, serve # noqa: E402
FAKEBIN = ROOT / "tests" / "fakebin"
@pytest.fixture
def bao():
b = FakeBao()
srv, port = serve(b)
b.port = port
yield b
srv.shutdown()
@pytest.fixture
def tenancy(bao):
t = FakeTenancy(bao)
srv, port = serve(t)
t.port = port
t.server = srv
yield t
try:
srv.shutdown()
except Exception:
pass
@pytest.fixture
def fake_docker(tmp_path, monkeypatch):
log = tmp_path / "docker.log"
state = tmp_path / "docker-state.json"
state.write_text(
json.dumps(
{
"ps": [{"Name": "see", "State": "running", "Health": "healthy"}],
"free_root": str(tmp_path),
"env_file_check": True,
}
)
)
monkeypatch.setenv("PATH", f"{FAKEBIN}:{os.environ['PATH']}")
monkeypatch.setenv("FAKE_DOCKER_LOG", str(log))
monkeypatch.setenv("FAKE_DOCKER_STATE", str(state))
class FD:
def calls(self):
if not log.exists():
return []
return [json.loads(line)["argv"] for line in log.read_text().splitlines()]
def subcommands(self):
out = []
for argv in self.calls():
if argv[:1] == ["compose"]:
rest = argv[1:]
while rest and rest[0].startswith("--"):
rest = rest[2:]
out.append("compose " + (rest[0] if rest else ""))
else:
out.append(" ".join(argv[:2]))
return out
def set(self, **kw):
cur = json.loads(state.read_text())
cur.update(kw)
state.write_text(json.dumps(cur))
return FD()
@pytest.fixture
def cfg(tmp_path, tenancy, bao, fake_docker):
REDACTOR.forget_all()
state_dir = tmp_path / "state"
etc = tmp_path / "etc"
etc.mkdir()
c = configmod.from_dict(
{
"env_id": ENV,
"site": "cbs",
"transport": "system",
"tenancy": {"host": "127.0.0.1", "port": tenancy.port, "scheme": "http"},
"bao": {"addr": f"http://127.0.0.1:{bao.port}", "ca_bundle": None},
"state_dir": str(state_dir),
"bootstrap_path": str(etc / "bootstrap.jwt"),
"interval_s": 60,
"healthy_timeout_s": 3,
"disk": {"factor": 1.5, "headroom_bytes": 1024},
},
path=str(etc / "config.yaml"),
)
return c
@pytest.fixture
def bootstrapped(cfg, bao):
"""A box with the kit's bootstrap grant on disk (first tick)."""
Path(cfg.bootstrap_path).write_text(bao.grant() + "\n")
return cfg
+68
View File
@@ -0,0 +1,68 @@
#!/usr/bin/env python3
"""Fake `docker` for the hermetic tests. Records every invocation to $FAKE_DOCKER_LOG (one JSON
line per call) and answers from $FAKE_DOCKER_STATE (a JSON file the test writes):
{"ps": [{"Name": "see", "State": "running", "Health": "healthy"}], # compose ps output
"fail": ["pull"], # sub-commands that must exit 1
"free_root": "/tmp"} # DockerRootDir for `docker info`
"""
import json
import os
import sys
def main() -> int:
argv = sys.argv[1:]
state = {}
sp = os.environ.get("FAKE_DOCKER_STATE")
if sp and os.path.exists(sp):
with open(sp) as fh:
state = json.load(fh)
lp = os.environ.get("FAKE_DOCKER_LOG")
if lp:
with open(lp, "a") as fh:
fh.write(json.dumps({"argv": argv, "cwd": os.getcwd()}) + "\n")
fail = set(state.get("fail", []))
if argv[:1] == ["version"]:
print("28.3.0")
return 0
if argv[:1] == ["info"]:
print(state.get("free_root", "/"))
return 0
if argv[:2] == ["image", "prune"]:
return 0
if argv[:1] == ["compose"]:
# strip global compose flags
rest = argv[1:]
while rest and rest[0].startswith("--"):
rest = rest[2:]
sub = rest[0] if rest else ""
if sub == "version":
print("2.32.0")
return 0
if sub in fail:
print(f"fake docker: {sub} failed", file=sys.stderr)
return 1
if sub == "ps":
for row in state.get("ps", []):
print(json.dumps(row))
return 0
if sub == "logs":
print("fake compose logs")
return 0
if sub == "up" and state.get("env_file_check"):
# prove the .env is complete and 0600
d = os.getcwd()
env = os.path.join(d, ".env")
st = os.stat(env)
if st.st_mode & 0o077:
print("fake docker: .env is not 0600", file=sys.stderr)
return 1
return 0
print(f"fake docker: unknown {argv}", file=sys.stderr)
return 2
if __name__ == "__main__":
sys.exit(main())
+374
View File
@@ -0,0 +1,374 @@
"""Fake monky-tenancy agent endpoint + fake OpenBao, as real HTTP servers on 127.0.0.1.
They implement exactly the shapes of record (ADR-0028 + Gate 1 v2): the lease is a
`{login_jwt, ttl_s, mount, role, vault}` deploy grant; OpenBao's `auth/jwt-tenancy/login`
accepts any grant listed in `bao.grants` and pins the token's meta to that grant's env_id."""
from __future__ import annotations
import io
import json
import tarfile
import threading
import time
import uuid
from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
from monky_deployd.bundle import bundle_sha
ENV = "env-qa-02"
COMPOSE = """services:
see:
image: harbor.tikali.net/monky/see-backend:${SEE_TAG:-develop}
environment:
POSTGRES_PASSWORD: ${POSTGRES_PASSWORD}
GEMINI_API_KEY: ${GEMINI_API_KEY}
SEE_ADMIN_TOKEN: ${SEE_ADMIN_TOKEN}
healthcheck:
test: ["CMD", "true"]
"""
def make_manifest(env=ENV, versions=None, prefix="monky"):
versions = versions or {}
entries = [
{
"var": "GEMINI_API_KEY",
"path": f"{prefix}/{env}/see/gemini_api_key",
"kind": "supplied",
"version": versions.get("gemini_api_key"),
},
{
"var": "POSTGRES_PASSWORD",
"path": f"{prefix}/{env}/see/pg_password",
"kind": "generated",
"version": versions.get("pg_password", 1),
},
{
"var": "SEE_ADMIN_TOKEN",
"path": f"{prefix}/{env}/see/admin_token",
"kind": "generated",
"version": versions.get("admin_token", 1),
},
]
return {"vault_mode": "openbao", "entries": entries}
def env_template(manifest):
lines = ["# .env template — PLACEHOLDERS ONLY", ""]
for e in manifest["entries"]:
lines += [f"# [{e['kind']}] openbao: {e['path']}", f"{e['var']}=${{{e['var']}}}"]
return "\n".join(lines) + "\n"
def make_files(env=ENV, *, compose=COMPOSE, manifest=None, meta=None):
manifest = manifest or make_manifest(env)
files = {
"docker-compose.yml": compose,
".env.template": env_template(manifest),
"secrets.manifest.json": json.dumps(manifest, indent=2, sort_keys=True) + "\n",
}
bundle_json = {
"env_id": env,
"tier": env.split("-")[1] if env.startswith("env-") else "dev",
"site": "cbs",
"bundle": "see",
"target_class": "docker-host",
"renderer": "compose",
"images_policy": "develop",
"secrets_provider": "openbao",
"agent": {},
"files": sorted([*files, "bundle.json"]),
}
if meta:
bundle_json.update(meta)
files["bundle.json"] = json.dumps(bundle_json, indent=2, sort_keys=True) + "\n"
return files
def tar_bytes(files: dict[str, str]) -> bytes:
buf = io.BytesIO()
with tarfile.open(fileobj=buf, mode="w") as tar:
for name in sorted(files):
data = files[name].encode()
info = tarfile.TarInfo(name=name)
info.size = len(data)
info.mode = 0o644
tar.addfile(info, io.BytesIO(data))
return buf.getvalue()
class _Server(ThreadingHTTPServer):
daemon_threads = True
allow_reuse_address = True
class _Handler(BaseHTTPRequestHandler):
server_version = "fake/0"
protocol_version = "HTTP/1.1"
def log_message(self, *a): # quiet
pass
def _body(self):
n = int(self.headers.get("Content-Length") or 0)
raw = self.rfile.read(n) if n else b""
return json.loads(raw) if raw else {}
def _send(self, status, body=None, *, raw=None, headers=None, ctype="application/json"):
data = raw if raw is not None else (json.dumps(body).encode() if body is not None else b"")
self.send_response(status)
self.send_header("Content-Type", ctype)
self.send_header("Content-Length", str(len(data)))
for k, v in (headers or {}).items():
self.send_header(k, v)
self.end_headers()
self.wfile.write(data)
def do_GET(self):
self.server.app.handle(self, "GET", self.path, {})
def do_POST(self):
self.server.app.handle(self, "POST", self.path, self._body())
class FakeBao:
"""auth/jwt-tenancy/login, token lookup-self/renew-self/revoke-self, KV-v2 reads."""
def __init__(self, env=ENV):
self.env = env
self.grants: dict[str, str] = {} # jwt -> env_id
self.tokens: dict[str, dict] = {}
self.kv: dict[str, list[dict]] = {} # "<env>/see/<name>" -> versions [ {value} ]
self.calls: list[tuple] = []
self.login_fail = False
self.token_ttl = 86400
self.max_ttl = 30 * 86400
self.mount = "jwt-tenancy"
self.role = "see-env"
self.kv_mount = "monky"
self.seed_secrets(env)
def seed_secrets(self, env):
self.kv[f"{env}/see/gemini_api_key"] = [
{"value": "AIzaSy-FAKE-GEMINI-KEY-0001"},
{"value": "AIzaSy-FAKE-GEMINI-KEY-0002"},
]
self.kv[f"{env}/see/pg_password"] = [{"value": 'pg-s3cret-"quoted"$dollar'}]
self.kv[f"{env}/see/admin_token"] = [{"value": "adm1n-t0ken-abcdef0123456789"}]
def grant(self, env=None, jti=None) -> str:
jwt = "eyJhbGciOiJFUzI1NiJ9." + uuid.uuid4().hex + uuid.uuid4().hex + "." + uuid.uuid4().hex + uuid.uuid4().hex
self.grants[jwt] = json.dumps({"env_id": env or self.env, "jti": jti or uuid.uuid4().hex})
return jwt
def mint(self, env=None, *, age_s=0, ttl=None) -> str:
tok = "hvs." + uuid.uuid4().hex + uuid.uuid4().hex[:8]
self.tokens[tok] = {
"accessor": "acc-" + uuid.uuid4().hex[:12],
"env_id": env or self.env,
"grant_jti": uuid.uuid4().hex,
"creation_time": int(time.time()) - age_s,
"ttl": ttl if ttl is not None else self.token_ttl,
"renewable": True,
"revoked": False,
}
return tok
def _auth(self, h):
tok = h.headers.get("X-Vault-Token")
t = self.tokens.get(tok)
if not t or t["revoked"]:
return None, tok
return t, tok
def handle(self, h, method, path, body):
self.calls.append((method, path.split("?")[0]))
if method == "POST" and path == f"/v1/auth/{self.mount}/login":
if self.login_fail:
return h._send(400, {"errors": ["error validating token: expired"]})
if body.get("role") != self.role:
return h._send(400, {"errors": [f"role {body.get('role')!r} could not be found"]})
meta = self.grants.pop(body.get("jwt", ""), None) # single-use grant
if meta is None:
return h._send(400, {"errors": ["error validating token: unknown or already used grant"]})
m = json.loads(meta)
tok = self.mint(m["env_id"])
self.tokens[tok]["grant_jti"] = m["jti"]
t = self.tokens[tok]
return h._send(
200,
{
"auth": {
"client_token": tok,
"accessor": t["accessor"],
"lease_duration": t["ttl"],
"renewable": True,
"token_policies": ["see-env"],
"metadata": {"env_id": m["env_id"], "grant_jti": m["jti"], "role": self.role},
}
},
)
t, tok = self._auth(h)
if t is None:
return h._send(403, {"errors": ["permission denied"]})
if method == "GET" and path == "/v1/auth/token/lookup-self":
return h._send(
200,
{
"data": {
"accessor": t["accessor"],
"creation_time": t["creation_time"],
"ttl": t["ttl"],
"renewable": t["renewable"],
"explicit_max_ttl": self.max_ttl,
"meta": {"env_id": t["env_id"], "grant_jti": t["grant_jti"]},
}
},
)
if method == "POST" and path == "/v1/auth/token/renew-self":
age = int(time.time()) - t["creation_time"]
t["ttl"] = max(0, min(self.token_ttl, self.max_ttl - age))
return h._send(200, {"auth": {"client_token": tok, "lease_duration": t["ttl"], "renewable": True}})
if method == "POST" and path == "/v1/auth/token/revoke-self":
t["revoked"] = True
return h._send(204)
if method == "GET" and path.startswith(f"/v1/{self.kv_mount}/data/"):
p, _, q = path.partition("?")
key = p[len(f"/v1/{self.kv_mount}/data/") :]
env = key.split("/")[0]
if env != t["env_id"]:
return h._send(403, {"errors": ["1 error occurred:\n\t* permission denied\n\n"]})
versions = self.kv.get(key)
if not versions:
return h._send(404, {"errors": []})
want = None
for part in q.split("&"):
if part.startswith("version="):
want = int(part.split("=", 1)[1])
idx = (want or len(versions)) - 1
if idx < 0 or idx >= len(versions):
return h._send(404, {"errors": []})
return h._send(200, {"data": {"data": versions[idx], "metadata": {"version": idx + 1}}})
return h._send(404, {"errors": [f"no handler for {method} {path}"]})
class FakeTenancy:
"""The agent entrypoint. `bao` validates bearers (a token for another env -> AGENT_ENV_MISMATCH)."""
def __init__(self, bao: FakeBao, env=ENV):
self.bao = bao
self.env = env
self.files = make_files(env)
self.desired_sha = bundle_sha(self.files)
self.action = "apply"
self.purge_volumes = False
self.checkins: list[dict] = []
self.reports: list[dict] = []
self.leases: list[dict] = []
self.lease_shape = "jwt" # or "approle" to simulate the un-migrated tenancy
self.lease_limit = 5
self.superseded_jtis: set[str] = set()
self.down = False # network down -> connection refused simulated by test stopping server
self.kv_mount = "monky"
def set_files(self, files):
self.files = files
self.desired_sha = bundle_sha(files)
def _principal(self, h):
auth = h.headers.get("Authorization", "")
tok = auth.split(" ", 1)[1] if auth.lower().startswith("bearer ") else None
t = self.bao.tokens.get(tok or "")
if not t or t["revoked"]:
return None, ("AGENT_UNAUTHENTICATED", "token unknown, expired or revoked")
if t["grant_jti"] in self.superseded_jtis:
return None, ("AGENT_UNAUTHENTICATED", "grant superseded")
return t, None
def _err(self, h, status, code, detail, headers=None):
return h._send(status, {"code": code, "detail": detail}, headers=headers)
def handle(self, h, method, path, body):
t, err = self._principal(h)
if err:
return self._err(h, 401, *err)
env_in = body.get("env_id") if body else None
if path.startswith("/v1/agent/bundle/"):
env_in = path.split("/")[4]
if env_in and env_in != t["env_id"]:
return self._err(h, 403, "AGENT_ENV_MISMATCH", f"token is pinned to {t['env_id']}, not {env_in}")
if method == "POST" and path == "/v1/agent/checkin":
self.checkins.append(body)
action = self.action
if action == "apply" and body.get("applied_sha") == self.desired_sha:
action = "none"
return h._send(
200,
{
"env_id": self.env,
"desired_sha": None if self.action == "down" else self.desired_sha,
"action": action,
"purge_volumes": self.purge_volumes if action == "down" else False,
"bundle_url": f"/v1/agent/bundle/{self.env}/{self.desired_sha}",
"checkin_interval_s": 60,
"vault": {
"addr": "https://bao.cbs.tikali.net:8200",
"mount": self.kv_mount,
"prefix": f"{self.env}/see",
},
},
)
if method == "GET" and path.startswith("/v1/agent/bundle/"):
sha = path.rsplit("/", 1)[1]
if sha != self.desired_sha:
return self._err(h, 404, "BUNDLE_NOT_FOUND", f"no published bundle {sha}")
return h._send(
200,
raw=tar_bytes(self.files),
ctype="application/x-tar",
headers={"Cache-Control": "no-store", "X-Bundle-Sha": sha},
)
if method == "POST" and path == "/v1/agent/lease":
if len(self.leases) >= self.lease_limit:
return self._err(h, 429, "LEASE_RATE_LIMITED", "5 leases per hour", headers={"Retry-After": "600"})
self.leases.append(body)
if self.lease_shape == "approle":
return h._send(
200,
{
"env_id": self.env,
"wrapping_token": "hvs.wrap-legacy",
"wrap_ttl_s": 3600,
"role_id": "role-id-see-env",
"vault": {},
},
)
jwt = self.bao.grant(self.env)
return h._send(
200,
{
"env_id": self.env,
"login_jwt": jwt,
"ttl_s": 3600,
"mount": self.bao.mount,
"role": self.bao.role,
"vault": {"addr": "https://bao.cbs.tikali.net:8200", "mount": self.kv_mount},
},
headers={"Cache-Control": "no-store"},
)
if method == "POST" and path == "/v1/agent/report":
self.reports.append(body)
if body.get("result") not in ("applied", "failed", "down"):
return self._err(h, 400, "BAD_REQUEST", "unknown result")
return h._send(200, {"env_id": self.env, "accepted": True, "result": body["result"]})
return self._err(h, 404, "NOT_FOUND", path)
def serve(app):
srv = _Server(("127.0.0.1", 0), _Handler)
srv.app = app
th = threading.Thread(target=srv.serve_forever, daemon=True)
th.start()
return srv, srv.server_address[1]
+321
View File
@@ -0,0 +1,321 @@
"""End-to-end ticks against the fake tenancy + fake OpenBao + fake docker."""
from __future__ import annotations
import json
import os
import time
from pathlib import Path
from monky_deployd import state as statemod
from monky_deployd.agent import EX_ENV_MISMATCH, EX_FAIL, EX_OK, EX_TEMPFAIL, Agent
from monky_deployd.bundle import bundle_sha
from tests.fakes import ENV, make_files, make_manifest
def tick(cfg, **kw):
return Agent(cfg, **kw).run_once()
def test_first_tick_bootstraps_applies_and_reports(bootstrapped, tenancy, bao, fake_docker):
cfg = bootstrapped
assert tick(cfg) == EX_OK
# bootstrap grant consumed, token persisted 0600
assert not Path(cfg.bootstrap_path).exists()
assert cfg.token_path.exists() and (cfg.token_path.stat().st_mode & 0o777) == 0o600
# protocol: checkin -> bundle -> lease -> report applied
assert tenancy.checkins[0]["env_id"] == ENV and tenancy.checkins[0]["agent_version"] == "0.1.0"
assert tenancy.checkins[0]["host"]["docker"] == "28.3.0"
assert len(tenancy.leases) == 1 and tenancy.leases[0]["reason"] == "apply"
assert [r["result"] for r in tenancy.reports] == ["applied"]
assert tenancy.reports[0]["sha"] == tenancy.desired_sha
assert tenancy.reports[0]["containers"] == [{"name": "see", "state": "running", "health": "healthy"}]
# OpenBao: login twice (bootstrap + lease), three KV reads pinned to versions
kv_calls = [p for m, p in bao.calls if "/data/" in p]
assert len(kv_calls) == 3 and all(f"/data/{ENV}/see/" in p for p in kv_calls)
# the lease token became the bearer; the bootstrap token was revoked
tokens = list(bao.tokens.values())
assert sum(1 for t in tokens if t["revoked"]) == 1
# files: .env 0600, complete, values never in logs; release promoted
current = Path(cfg.deploy_dir) / "current"
env = (current / ".env").read_text()
assert (current / ".env").stat().st_mode & 0o077 == 0
assert "GEMINI_API_KEY=AIzaSy-FAKE-GEMINI-KEY-0002" in env # latest version (manifest version None)
assert 'POSTGRES_PASSWORD="pg-s3cret-\\"quoted\\"$$dollar"' in env
assert "${" not in env.replace("${", "") or "=${" not in env
assert (current / "SHA256").read_text().strip() == tenancy.desired_sha
# compose: pull then up then ps
subs = fake_docker.subcommands()
assert "compose pull" in subs and "compose up" in subs and subs.index("compose pull") < subs.index("compose up")
# state
st = statemod.load(cfg.state_path, ENV)
assert st.applied_sha == tenancy.desired_sha and st.last_result == "applied" and st.token.source == "lease"
# log tail never carries a value
tail = tenancy.reports[0]["log_tail"]
assert tail and "AIzaSy" not in tail and "pg-s3cret" not in tail and "hvs." not in tail
def test_second_tick_is_a_heartbeat(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
assert tick(cfg) == EX_OK
assert tick(cfg) == EX_OK
assert [r["result"] for r in tenancy.reports] == ["applied", "applied"]
assert len(tenancy.leases) == 1 # no lease for a heartbeat
assert tenancy.checkins[1]["applied_sha"] == tenancy.desired_sha
assert fake_docker.subcommands().count("compose pull") == 1
def test_none_but_unhealthy_reports_failed(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
assert tick(cfg) == EX_OK
fake_docker.set(ps=[{"Name": "see", "State": "exited", "ExitCode": 137, "Health": ""}])
assert tick(cfg) == EX_FAIL
assert tenancy.reports[-1]["result"] == "failed" and "see=exited" in tenancy.reports[-1]["detail"]
def test_rotation_changes_sha_and_reapplies(bootstrapped, tenancy, bao, fake_docker):
cfg = bootstrapped
assert tick(cfg) == EX_OK
old = tenancy.desired_sha
tenancy.set_files(make_files(manifest=make_manifest(versions={"gemini_api_key": 1})))
assert tenancy.desired_sha != old
assert tick(cfg) == EX_OK
env = (Path(cfg.deploy_dir) / "current" / ".env").read_text()
assert "GEMINI_API_KEY=AIzaSy-FAKE-GEMINI-KEY-0001" in env # pinned version 1
st = statemod.load(cfg.state_path, ENV)
assert st.history == [old, tenancy.desired_sha]
assert len(list((Path(cfg.deploy_dir) / "releases").iterdir())) == 2
def test_prune_keeps_only_current(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
assert tick(cfg) == EX_OK
tenancy.set_files(make_files(manifest=make_manifest(versions={"gemini_api_key": 1})))
assert tick(cfg, prune=True) == EX_OK
assert [d.name for d in (Path(cfg.deploy_dir) / "releases").iterdir()] == [tenancy.desired_sha]
assert "image prune" in fake_docker.subcommands()
def test_rollback_refused_unless_allowed(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
first = make_files()
second = make_files(manifest=make_manifest(versions={"gemini_api_key": 1}))
assert tick(cfg) == EX_OK
tenancy.set_files(second)
assert tick(cfg) == EX_OK
tenancy.set_files(first)
assert tick(cfg) == EX_FAIL
assert tenancy.reports[-1]["result"] == "failed" and "ROLLBACK_REFUSED" in tenancy.reports[-1]["detail"]
assert len(tenancy.leases) == 2 # a refused bundle never leases
tenancy.set_files(make_files(meta={"agent": {"allow_rollback": True}}))
assert tick(cfg) == EX_OK
def test_down_removes_orphans_and_purges_only_when_allowed(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
assert tick(cfg) == EX_OK
tenancy.action, tenancy.purge_volumes = "down", True
assert tick(cfg) == EX_OK
down = [a for a in fake_docker.calls() if "down" in a][-1]
assert "--remove-orphans" in down and "-v" in down
assert tenancy.reports[-1] == {
"env_id": ENV,
"sha": None,
"result": "down",
"log_tail": tenancy.reports[-1]["log_tail"],
"containers": None,
}
assert statemod.load(cfg.state_path, ENV).applied_sha is None
assert not (Path(cfg.deploy_dir) / "current").exists()
def test_down_on_prod_never_purges(cfg, tenancy, bao, fake_docker):
cfg.prod = True
cfg.token_path.parent.mkdir(parents=True)
statemod.write_token(cfg.token_path, bao.mint())
tenancy.action, tenancy.purge_volumes = "down", True
assert tick(cfg) == EX_OK
down = [a for a in fake_docker.calls() if "down" in a][-1]
assert "-v" not in down and "--remove-orphans" in down
def test_env_mismatch_exits_78_and_writes_nothing(cfg, tenancy, bao, fake_docker):
cfg.token_path.parent.mkdir(parents=True)
statemod.write_token(cfg.token_path, bao.mint(env="env-dev-01"))
assert tick(cfg) == EX_ENV_MISMATCH
assert not (Path(cfg.deploy_dir) / "current").exists()
assert tenancy.reports == [] and tenancy.leases == []
assert "AGENT_ENV_MISMATCH" in statemod.load(cfg.state_path, ENV).last_error
def test_superseded_grant_drops_token_and_rebootstraps_if_grant_present(cfg, tenancy, bao, fake_docker):
cfg.token_path.parent.mkdir(parents=True)
old = bao.mint()
statemod.write_token(cfg.token_path, old)
tenancy.superseded_jtis.add(bao.tokens[old]["grant_jti"])
# no bootstrap grant -> exit 1, token removed, operator told to re-run the kit
assert tick(cfg) == EX_FAIL
assert not cfg.token_path.exists()
assert "AGENT_UNAUTHENTICATED" in statemod.load(cfg.state_path, ENV).last_error
# with a fresh kit grant on disk the same tick recovers
Path(cfg.bootstrap_path).write_text(bao.grant())
assert tick(cfg) == EX_OK
assert tenancy.reports[-1]["result"] == "applied"
def test_no_credentials_is_a_clear_exit_1(cfg, tenancy, fake_docker):
assert tick(cfg) == EX_FAIL
assert "install kit" in statemod.load(cfg.state_path, ENV).last_error
assert tenancy.checkins == []
def test_network_down_is_75_or_0_in_laptop_mode(cfg, tenancy, bao, fake_docker):
cfg.token_path.parent.mkdir(parents=True)
statemod.write_token(cfg.token_path, bao.mint())
tenancy.server.shutdown()
tenancy.server.server_close()
assert tick(cfg) == EX_TEMPFAIL
cfg.laptop_mode = True
assert tick(cfg) == EX_OK
assert "network" in statemod.load(cfg.state_path, ENV).last_error
def test_unresolved_var_is_env_incomplete_names_only(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
tenancy.set_files(
make_files(
compose="services:\n x:\n image: i\n environment:\n A: ${SOMETHING_MISSING}\n B: ${POSTGRES_PASSWORD}\n"
)
)
assert tick(cfg) == EX_FAIL
rep = tenancy.reports[-1]
assert rep["result"] == "failed" and "ENV_INCOMPLETE" in rep["detail"] and "SOMETHING_MISSING" in rep["detail"]
assert tenancy.leases == [] and "compose up" not in fake_docker.subcommands()
def test_privileged_refused_unless_bundle_allows(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
priv = "services:\n x:\n image: i\n privileged: true\n environment:\n B: ${POSTGRES_PASSWORD}\n A: ${GEMINI_API_KEY}\n C: ${SEE_ADMIN_TOKEN}\n"
tenancy.set_files(make_files(compose=priv))
assert tick(cfg) == EX_FAIL
assert "PRIVILEGED_REFUSED" in tenancy.reports[-1]["detail"]
tenancy.set_files(make_files(compose=priv, meta={"allow_privileged": True}))
assert tick(cfg) == EX_OK
def test_manifest_path_outside_env_refused_before_any_read(bootstrapped, tenancy, bao, fake_docker):
cfg = bootstrapped
tenancy.set_files(make_files(manifest=make_manifest(env="env-dev-01")))
assert tick(cfg) == EX_FAIL
assert "outside" in tenancy.reports[-1]["detail"]
assert not [p for m, p in bao.calls if "/data/" in p]
def test_bundle_env_mismatch_refused(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
tenancy.set_files(make_files(meta={"env_id": "env-dev-01"}))
assert tick(cfg) == EX_FAIL
assert "BUNDLE_ENV_MISMATCH" in tenancy.reports[-1]["detail"]
def test_sha_mismatch_refused(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
files = make_files()
tenancy.files = files
tenancy.desired_sha = "0" * 64 # tenancy claims a sha the tar does not hash to
assert tick(cfg) == EX_FAIL
assert "BUNDLE_SHA_MISMATCH" in tenancy.reports[-1]["detail"]
assert bundle_sha(files) != tenancy.desired_sha
def test_disk_refusal(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
tenancy.set_files(make_files(meta={"agent": {"disk_need_bytes": 10**18}}))
assert tick(cfg) == EX_FAIL
assert "DISK_INSUFFICIENT" in tenancy.reports[-1]["detail"]
assert tenancy.leases == []
def test_legacy_approle_lease_is_refused_loudly(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
tenancy.lease_shape = "approle"
assert tick(cfg) == EX_FAIL
rep = tenancy.reports[-1]
assert rep["result"] == "failed" and "LEASE_SHAPE" in rep["detail"] and "jwt-tenancy" in rep["detail"]
assert "compose up" not in fake_docker.subcommands()
def test_lease_rate_limited_is_temporary(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
tenancy.lease_limit = 0
assert tick(cfg) == EX_TEMPFAIL
def test_unhealthy_after_up_reports_failed_with_compose_logs(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
fake_docker.set(ps=[{"Name": "see", "State": "running", "Health": "starting"}])
assert tick(cfg) == EX_FAIL
rep = tenancy.reports[-1]
assert rep["result"] == "failed" and "fake compose logs" in rep["log_tail"]
assert statemod.load(cfg.state_path, ENV).applied_sha is None
def test_compose_pull_failure_reports_failed(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
fake_docker.set(fail=["pull"])
assert tick(cfg) == EX_FAIL
assert tenancy.reports[-1]["result"] == "failed" and "pull" in tenancy.reports[-1]["detail"]
def test_token_renew_and_release_before_max_ttl(bootstrapped, tenancy, bao, fake_docker):
cfg = bootstrapped
assert tick(cfg) == EX_OK
tok = cfg.token_path.read_text()
# low TTL, renewable -> renew-self
bao.tokens[tok]["ttl"] = 100
assert tick(cfg) == EX_OK
assert ("POST", "/v1/auth/token/renew-self") in bao.calls and bao.tokens[tok]["ttl"] == bao.token_ttl
# near max TTL -> re-lease (a new token replaces the old, which is revoked)
bao.tokens[tok]["creation_time"] = int(time.time()) - (bao.max_ttl - 3600)
st = statemod.load(cfg.state_path, ENV)
st.token.issued_at = time.time() - 7200
statemod.save(cfg.state_path, st)
assert tick(cfg) == EX_OK
assert tenancy.leases[-1]["reason"] == "renew"
new = cfg.token_path.read_text()
assert new != tok and bao.tokens[tok]["revoked"] is True and bao.tokens[new]["revoked"] is False
def test_dead_stored_token_is_dropped_by_upkeep(bootstrapped, tenancy, bao, fake_docker):
cfg = bootstrapped
assert tick(cfg) == EX_OK
tok = cfg.token_path.read_text()
bao.tokens[tok]["revoked"] = True
tenancy.action = "none"
# tenancy refuses the revoked bearer -> 401 path (no grant on disk) -> token dropped
assert tick(cfg) == EX_FAIL
assert not cfg.token_path.exists()
def test_lock_prevents_overlap(bootstrapped, fake_docker):
cfg = bootstrapped
lock = statemod.Lock(cfg.lock_path)
assert lock.acquire()
try:
assert tick(cfg) == EX_OK # skipped quietly
assert Path(cfg.bootstrap_path).exists() # nothing happened
finally:
lock.release()
def test_state_for_another_env_is_reset(tmp_path):
p = tmp_path / "state.json"
p.write_text(json.dumps({"env_id": "env-dev-01", "applied_sha": "x"}))
st = statemod.load(p, ENV)
assert st.env_id == ENV and st.applied_sha is None and "env-dev-01" in st.last_error
def test_write_private_mode(tmp_path):
p = tmp_path / "d" / "f"
statemod.write_private(p, b"x")
assert oct(p.stat().st_mode & 0o777) == "0o600" and not any(n.startswith(".f.") for n in os.listdir(p.parent))
+81
View File
@@ -0,0 +1,81 @@
import pytest
from monky_deployd import bundle as b
from monky_deployd.bao import ManifestPathError, kv_data_path
from tests.fakes import ENV, make_files, make_manifest, tar_bytes
def test_parse_and_sha_match_tenancy_formula():
files = make_files()
bd = b.parse(tar_bytes(files))
assert bd.sha == b.bundle_sha(files) and len(bd.sha) == 64
assert bd.compose_name == "docker-compose.yml"
assert bd.env_id == ENV and bd.tier == "qa"
assert b.manifest_vars(bd) == {"GEMINI_API_KEY", "POSTGRES_PASSWORD", "SEE_ADMIN_TOKEN"}
assert b.unresolved_vars(bd, b.manifest_vars(bd)) == []
def test_unresolved_vars_names_only_defaults_resolve():
files = make_files(
compose="services:\n x:\n image: i:${TAG:-dev}\n environment:\n A: ${NOT_IN_MANIFEST}\n B: ${POSTGRES_PASSWORD}\n"
)
bd = b.parse(tar_bytes(files))
assert b.unresolved_vars(bd, b.manifest_vars(bd)) == ["NOT_IN_MANIFEST"]
def test_privileged_and_host_network_detected():
files = make_files(
compose="services:\n x:\n image: i\n privileged: true\n network_mode: host\n cap_add:\n - SYS_ADMIN\n"
)
bd = b.parse(tar_bytes(files))
assert b.privileged_findings(bd) == ["privileged: true", "network_mode: host", "cap_add SYS_ADMIN/ALL"]
assert bd.flag("allow_privileged") is False
bd2 = b.parse(tar_bytes(make_files(meta={"agent": {"allow_privileged": True}})))
assert bd2.flag("allow_privileged") is True
def test_manifest_with_value_is_refused():
m = make_manifest()
m["entries"][0]["value"] = "leaked"
with pytest.raises(b.BundleError, match="carries a value"):
b.parse(tar_bytes(make_files(manifest=m)))
def test_unsafe_members_refused():
import io
import tarfile
buf = io.BytesIO()
with tarfile.open(fileobj=buf, mode="w") as tar:
info = tarfile.TarInfo(name="../etc/passwd")
info.size = 1
tar.addfile(info, io.BytesIO(b"x"))
with pytest.raises(b.BundleError, match="unsafe"):
b.parse(buf.getvalue())
def test_render_env_quotes_the_compose_way():
tmpl = "# c\nA=${A}\nB=${B}\nC=${C}\nKEEP=${KEEP}\n"
out = b.render_env(tmpl, {"A": "plain-value.1", "B": 'has "quote" and $dollar', "C": "multi\nline"})
assert 'A=plain-value.1\nB="has \\"quote\\" and $$dollar"\nC="multi\\nline"\nKEEP=${KEEP}\n' in out
assert b.referenced_vars(out) == {"KEEP"}
def test_kv_paths_are_pinned_to_the_env():
assert kv_data_path("monky", f"monky/{ENV}/see/pg_password", ENV) == f"/v1/monky/data/{ENV}/see/pg_password"
assert kv_data_path("monky", f"monky/data/{ENV}/see/x", ENV) == f"/v1/monky/data/{ENV}/see/x"
for bad in (
"monky/env-dev-01/see/x",
"monky/companies/c1/ai/gemini",
"other/env-qa-02/see/x",
f"monky/{ENV}/zitadel/x",
f"monky/{ENV}/see/../x",
):
with pytest.raises(ManifestPathError):
kv_data_path("monky", bad, ENV)
def test_disk_need_bytes_spellings():
assert b.parse(tar_bytes(make_files(meta={"agent": {"disk_need_bytes": 5}}))).disk_need_bytes == 5
assert b.parse(tar_bytes(make_files(meta={"disk": {"need_bytes": 7}}))).disk_need_bytes == 7
assert b.parse(tar_bytes(make_files())).disk_need_bytes == 0
+102
View File
@@ -0,0 +1,102 @@
import json
from pathlib import Path
from monky_deployd import cli
from monky_deployd.agent import EX_OK
def write_cfg(cfg):
lines = [
f"env_id: {cfg.env_id}",
f"site: {cfg.site}",
"transport: system",
"tenancy:",
f" host: {cfg.tenancy.host}",
f" port: {cfg.tenancy.port}",
" scheme: http",
"bao:",
f" addr: {cfg.bao.addr}",
" ca_bundle: none",
f"state_dir: {cfg.state_dir}",
f"bootstrap_path: {cfg.bootstrap_path}",
f"deploy_dir: {cfg.deploy_dir}",
"healthy_timeout_s: 3",
]
Path(cfg.path).write_text("\n".join(lines) + "\n")
return cfg.path
def test_run_once_then_status(bootstrapped, tenancy, fake_docker, capsys):
path = write_cfg(bootstrapped)
assert cli.main(["-c", path, "run", "--once"]) == EX_OK
assert cli.main(["-c", path, "status", "--json"]) == EX_OK
out = json.loads(capsys.readouterr().out)
assert out["in_sync"] is True and out["applied_sha"] == tenancy.desired_sha and out["token_present"] is True
assert out["healthy"] is True and out["containers"][0]["name"] == "see"
assert cli.main(["-c", path, "status"]) == EX_OK
text = capsys.readouterr().out
assert "in sync" in text and "healthy" in text and "hvs." not in text
def test_bootstrap_command(bootstrapped, bao, capsys):
cfg = bootstrapped
path = write_cfg(cfg)
assert cli.main(["-c", path, "bootstrap"]) == EX_OK
assert cfg.token_path.exists() and not Path(cfg.bootstrap_path).exists()
assert cli.main(["-c", path, "bootstrap"]) == EX_OK
assert "already bootstrapped" in capsys.readouterr().out
def test_version_and_bad_config(capsys, tmp_path):
assert cli.main(["version"]) == 0
assert capsys.readouterr().out.strip() == "0.1.0"
bad = tmp_path / "c.yaml"
bad.write_text("env_id: nope\nsite: cbs\n")
assert cli.main(["-c", str(bad), "status"]) == 78
assert cli.main(["-c", str(tmp_path / "missing.yaml"), "status"]) == 1
def test_sdk_transport_uses_openziti_monkeypatch(monkeypatch, tmp_path):
"""The sdk transport loads the identity once and dials inside openziti.monkeypatch()."""
import contextlib
import socket
import sys
import types
calls = []
fake = types.ModuleType("openziti")
fake.load = lambda p: calls.append(("load", p)) or object()
@contextlib.contextmanager
def mp():
calls.append(("monkeypatch",))
yield
fake.monkeypatch = mp
monkeypatch.setitem(sys.modules, "openziti", fake)
from monky_deployd.transport import SdkTransport
srv = socket.socket()
srv.bind(("127.0.0.1", 0))
srv.listen(1)
t = SdkTransport(str(tmp_path / "id.json"))
s = t.connect("127.0.0.1", srv.getsockname()[1], 2)
s.close()
srv.close()
assert calls == [("load", str(tmp_path / "id.json")), ("monkeypatch",)]
assert "sdk(identity=" in t.describe()
def test_proxy_transport_refuses_unmapped_hosts():
from monky_deployd.config import from_dict
from monky_deployd.transport import TransportError, build
cfg = from_dict({"env_id": "env-dev-06", "site": "cbs", "transport": "proxy"})
t = build(cfg)
assert t.mapping[("monky.tenancy.deploy", 8081)] == ("127.0.0.1", 18443)
assert t.mapping[("bao.cbs.tikali.net", 8200)] == ("127.0.0.1", 18200)
try:
t.connect("example.com", 443, 1)
raise AssertionError("unmapped host must be refused")
except TransportError:
pass
+79
View File
@@ -0,0 +1,79 @@
import pytest
from monky_deployd import config as c
KIT = """\
# written by the install kit
env_id: env-qa-02
site: cbs
transport: sdk
identity: /opt/openziti/etc/identities/monky-host.env-qa-02.json
tenancy:
service: monky.tenancy.deploy
base_url: http://monky.tenancy.deploy:8081
bao:
service: openbao
addr: https://bao.cbs.tikali.net:8200 # intercept, not public DNS
ca_bundle: /etc/monky-deployd/openbao-ca.pem
mount: jwt-tenancy
role: see-env
kv_mount: monky
interval_s: 60
laptop_mode: false
volumes_on_absent: keep
"""
def test_yaml_subset_parses_nested_maps_and_types():
d = c.parse_yaml_subset(KIT)
assert d["env_id"] == "env-qa-02"
assert d["tenancy"]["base_url"] == "http://monky.tenancy.deploy:8081"
assert d["bao"]["addr"] == "https://bao.cbs.tikali.net:8200"
assert d["interval_s"] == 60 and d["laptop_mode"] is False
def test_yaml_subset_lists_quotes_and_comments():
d = c.parse_yaml_subset("a: \"x # not a comment\"\nb: 'q'\nlist:\n - one\n - 2\nn: ~\n")
assert d == {"a": "x # not a comment", "b": "q", "list": ["one", 2], "n": None}
def test_yaml_subset_refuses_flow_style_and_tabs():
with pytest.raises(c.ConfigError):
c.parse_yaml_subset("a: [1, 2]\n")
with pytest.raises(c.ConfigError):
c.parse_yaml_subset("a:\n\tb: 1\n")
def test_config_defaults_and_derivations():
cfg = c.from_dict(c.parse_yaml_subset(KIT))
assert cfg.tenancy.host == "monky.tenancy.deploy" and cfg.tenancy.port == 8081 and cfg.tenancy.scheme == "http"
assert cfg.bao_url == ("https", "bao.cbs.tikali.net", 8200)
assert cfg.deploy_dir == "/var/lib/monky-deployd/env-qa-02"
assert cfg.compose_project == "monky-env-qa-02"
assert str(cfg.token_path) == "/var/lib/monky-deployd/bao.token"
assert cfg.is_prod is False
assert cfg.bao.mount == "jwt-tenancy" and cfg.bao.role == "see-env"
def test_config_prod_detection_and_legacy_ids():
assert c.from_dict({"env_id": "env-prod-01", "site": "pdx"}).is_prod is True
assert c.from_dict({"env_id": "prod-cedar", "site": "cbs"}).is_prod is True
assert c.from_dict({"env_id": "dev-env-2", "site": "cbs"}).is_prod is False
with pytest.raises(c.ConfigError):
c.from_dict({"env_id": "dev-env-1", "site": "cbs"}) # retired 2026-09-04
def test_config_rejects_approle_and_unknown_keys():
with pytest.raises(c.ConfigError, match="approle"):
c.from_dict({"env_id": "env-dev-06", "site": "cbs", "bao": {"approle": {"path": "approle"}}})
with pytest.raises(c.ConfigError, match="unknown key"):
c.from_dict({"env_id": "env-dev-06", "site": "cbs", "tenancy": {"nope": 1}})
with pytest.raises(c.ConfigError, match="transport"):
c.from_dict({"env_id": "env-dev-06", "site": "cbs", "transport": "carrier-pigeon"})
with pytest.raises(c.ConfigError, match="site"):
c.from_dict({"env_id": "env-dev-06", "site": "sfo"})
def test_sdk_identity_defaults_to_host_identity():
cfg = c.from_dict({"env_id": "env-dev-07", "site": "cbs"})
assert cfg.identity == "/opt/openziti/etc/identities/monky-host.env-dev-07.json"
+27
View File
@@ -0,0 +1,27 @@
import logging
from monky_deployd.redact import MASK, Redactor
def test_redacts_token_shapes_and_registered_values():
r = Redactor()
r.add("pg-s3cret-value")
text = "tok hvs.CAESIJabcdefghijklmnopqrstuvwxyz0123456789 jwt eyJhbGciOiJFUzI1NiJ9.eyJzdWIiOiJhZ2VudCJ9.c2lnbmF0dXJlLXNpZw pw=pg-s3cret-value ok"
out = r.scrub(text)
assert "hvs." not in out and "eyJ" not in out and "pg-s3cret-value" not in out
assert out.count(MASK) == 3
assert r.scrub("Authorization: Bearer abc.def") == f"Authorization: Bearer {MASK}"
def test_filter_rewrites_records_in_place():
r = Redactor()
r.add("SUPERSECRET")
rec = logging.LogRecord("x", logging.INFO, "f", 1, "value=%s", ("SUPERSECRET",), None)
assert r.filter(rec) is True
assert rec.getMessage() == f"value={MASK}"
def test_short_values_are_not_registered():
r = Redactor()
r.add("ab")
assert r.scrub("ab is fine") == "ab is fine"