admin/felhom-agent
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

R-840: the config bundle — a signed agent_config_update brings a box's root-owned files (sudoers, wrappers, units)
gates / gates (push) Successful in 18s
Details

Browse Source 
felhom-os-apply gains mode 'bundle' (signed, verified by the wrapper itself against the root-owned
signers file — or, when that file is missing, only the installer's pinned key, which it then creates)
and --install-bundle (the installer's root entry). BUNDLE_FILES is the one table of paths; every check
(visudo, sh/bash -n, python, unit sections, RuntimeDirectory guard, nft -c, the route itself) runs
before the first write; a failed write or self-check puts every previous copy back. The trust root is
never a bundle path (R17). scripts/build-config-bundle.py builds it reproducibly; release-agent.sh
publishes it beside the binary. The agent reports the bundle record in system.config_bundle.
felhom-opsign signs agent_config_update. 43 wrapper tests (22 mutants red), Go executor tests.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
This commit is contained in:
authentik Default Admin
2026-10-04 19:36:58 +02:00
parent 0af1187e03
commit c9fa2e717b
15 changed files with 1554 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files
configs/felhom-os-apply
+500 -15
View File
@@ -93,6 +93,75 @@ DAEMON_JSON = "/etc/docker/daemon.json"
DOCKER_SOCKETS = ("/var/run/docker.sock", "/run/docker.sock")
CRASH_GUARD_STATE = "/var/lib/felhom-crash-guard/state.json"
# ---------- the config bundle (R-840, agent v0.143.0, `11` §5.4.2) ----------
# A signed `agent_config_update` job carries {agent_version, bundle_sha256}; the bundle is ONE JSON file built from this
# repo's configs/ at the agent tag (scripts/build-config-bundle.py) and published beside the binary. The installer
# installs the SAME bundle through this same code (--install-bundle, root only, never reachable through sudo), so a new
# box and an updated box cannot drift. This table is the ONLY set of paths a bundle may write — a signed bundle naming
# any other path is refused (R16) — and it is also the builder's list (one table). Each entry:
# dest: (source file under configs/, mode, check, policy)
# check: the content check BEFORE anything is written (R18): sudoers → visudo -cf; sh / bash → -n; python → compile;
# unit → a [Unit]/[Service]/[Timer] section; unit-nort → that, and never RuntimeDirectory= (the G1 incident);
# agent-unit → User=felhom-agent; dropin → a [Unit] section (SF-3); nft → nft -c -f; plain → none.
# policy: replace → always written when it differs; if-absent → only when the box has none (a setting the operator may
# have tuned); oob → only on a box with the OOB belt (/etc/felhom-sshd exists — the installer's --no-oob leaves none).
# Order matters: the sudoers files come LAST, so a referenced wrapper is in place before the line that allows it.
BUNDLE_FORMAT = 1
BUNDLE_OP = "agent_config_update"
BUNDLE_RECORD = "/etc/felhom/config-bundle.json" # what the box runs (0644 root; the non-root agent reports it)
BUNDLE_PREV_DIR = "/var/lib/felhom-os-apply/bundle-prev" # the previous copies of every file a bundle replaced
BUNDLE_KEEP = 3
BUNDLE_MAX_BYTES = 4 * 1024 * 1024
BUNDLE_RE = re.compile(r"^bundle-[0-9]+\.[0-9]+\.[0-9]+(-[0-9A-Za-z.]+)?\.json$")
OOB_DIR = "/etc/felhom-sshd"
BUNDLE_FILES = [
("/usr/local/sbin/felhom-mkfs-guarded", "felhom-mkfs-guarded.sh", 0o755, "bash", "replace"),
("/usr/local/sbin/felhom-selfupdate-guarded", "felhom-selfupdate-guarded", 0o755, "sh", "replace"),
("/usr/local/sbin/felhom-pbs-apply", "felhom-pbs-apply", 0o755, "bash", "replace"),
("/usr/local/sbin/felhom-backup-target-apply", "felhom-backup-target-apply", 0o755, "bash", "replace"),
("/usr/local/sbin/felhom-os-apply", "felhom-os-apply", 0o755, "python", "replace"),
("/usr/local/sbin/felhom-crash-guard", "felhom-crash-guard", 0o755, "python", "replace"),
("/etc/systemd/system/felhom-crash-guard.service", "felhom-crash-guard.service", 0o644, "unit", "replace"),
("/etc/systemd/system/felhom-crash-guard-check.service", "felhom-crash-guard-check.service", 0o644, "unit", "replace"),
("/etc/systemd/system/felhom-crash-guard-check.timer", "felhom-crash-guard-check.timer", 0o644, "unit", "replace"),
("/etc/felhom/crash-guard.conf", "crash-guard.conf", 0o644, "plain", "if-absent"),
("/etc/systemd/system/felhom-agent.service", "felhom-agent.service", 0o644, "agent-unit", "replace"),
("/etc/systemd/system/felhom-agent-rollback.service", "felhom-agent-rollback.service", 0o644, "unit", "replace"),
("/etc/systemd/system/felhom-agent.service.d/felhom-agent-limits.conf", "felhom-agent-limits.conf", 0o644, "dropin", "replace"),
("/usr/local/sbin/felhom-mgmt-watchdog", "felhom-mgmt-watchdog.sh", 0o755, "sh", "replace"),
("/etc/tmpfiles.d/felhom-privsep.conf", "felhom-privsep.tmpfiles", 0o644, "plain", "replace"),
("/etc/systemd/system/felhom-mgmt-watchdog.service", "felhom-mgmt-watchdog.service", 0o644, "unit-nort", "replace"),
("/etc/systemd/system/felhom-mgmt-watchdog.timer", "felhom-mgmt-watchdog.timer", 0o644, "unit-nort", "replace"),
("/etc/systemd/system/felhom-sshd.service", "felhom-sshd.service", 0o644, "unit-nort", "oob"),
("/etc/felhom-oob.nft", "felhom-oob.nft", 0o644, "nft", "oob"),
("/etc/systemd/system/felhom-oob-nft.service", "felhom-oob-nft.service", 0o644, "unit", "oob"),
("/etc/sudoers.d/felhom-op", "felhom-op.sudoers", 0o440, "sudoers", "oob"),
("/etc/sudoers.d/felhom-agent", "felhom-agent.sudoers", 0o440, "sudoers", "replace"),
]
BUNDLE_DESTS = {e[0]: e for e in BUNDLE_FILES}
# The trust root is never a bundle's to change (R17): who may sign is decided by these files, so no bundle may carry them.
TRUST_PATHS = (TRUST_FILE, TRUST_SIGNERS, BUNDLE_RECORD)
# When /etc/felhom/operator-signers is MISSING (a box installed before installer 1.30.0), the job is verified against —
# and the file is created with — exactly this key: the operational key felhom-host-install.sh pins
# (OPERATOR_KEY_OPERATIONAL_*). Never any other. Pinned equal to the installer by
# test_pinned_operator_key_equals_the_installers. Signer rotation is a separate, later act (`04` §3).
PINNED_OPERATOR_KEY_ID = "felhom-op-1"
PINNED_OPERATOR_KEY_LINE = ("ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIL8z0qCNgA3x2xxAB0Qj5ro8waFjGZ8Ta/sWB63tlLw+ felhom-op-1")
# The self-check (after install, before the record): the route itself must survive — a bundle whose sudoers no longer
# lets the agent call this wrapper would cut the box off from every later bundle.
SELF_CHECK_SUDO_LINE = "/usr/local/sbin/felhom-os-apply --plan /var/lib/felhom-agent/os/plan-*.json"
PINNED_SIGNERS = "<pinned operator key>" # verify_sig: check against the pinned key, not a file on the box
def pinned_signers_line():
"""The ssh allowed_signers line the installer writes (felhom-host-install.sh _write_slow_lane_trust), byte for byte."""
return f'{PINNED_OPERATOR_KEY_ID} namespaces="{SIG_NAMESPACE}" {PINNED_OPERATOR_KEY_LINE}\n'
def sha256_hex(data):
import hashlib
return hashlib.sha256(data).hexdigest()
class Refused(Exception):
def __init__(self, code, reason):
@@ -144,6 +213,10 @@ class Runner:
sp = os.path.join(d, "sig")
with open(sp, "w") as f:
f.write(sig)
if signers == PINNED_SIGNERS:
signers = os.path.join(d, "allowed_signers")
with open(signers, "w") as f:
f.write(pinned_signers_line())
p = subprocess.run(["ssh-keygen", "-Y", "verify", "-f", signers, "-I", key_id, "-n", namespace, "-s", sp],
input=blob, capture_output=True, timeout=30)
return p.returncode
@@ -170,6 +243,62 @@ class Runner:
except Exception:
pass
# ---------- host files, for the config bundle (R-840). Tests replace these with an in-memory tree. ----------
def read_bytes(self, path):
with open(path, "rb") as f:
return f.read()
def lexists(self, path):
return os.path.lexists(path)
def isdir(self, path):
return os.path.isdir(path)
def put_file(self, path, data, mode):
"""Atomic: a root-owned temp file beside the target (same filesystem), fsync, then rename over it."""
d = os.path.dirname(path)
os.makedirs(d, mode=0o755, exist_ok=True)
tmp = os.path.join(d, f".{os.path.basename(path)}.felhom-new.{os.getpid()}")
fd = os.open(tmp, os.O_WRONLY | os.O_CREAT | os.O_EXCL, 0o600)
try:
with os.fdopen(fd, "wb") as f:
f.write(data)
f.flush()
os.fchown(f.fileno(), 0, 0)
os.fchmod(f.fileno(), mode)
os.fsync(f.fileno())
os.replace(tmp, path)
except BaseException:
try:
os.remove(tmp)
except OSError:
pass
raise
def remove(self, path):
os.remove(path)
def list_dir(self, path):
try:
return sorted(os.listdir(path))
except OSError:
return []
def rmtree(self, path):
import shutil
shutil.rmtree(path, ignore_errors=True)
def check_content(self, kind, data):
"""Run an external syntax check on a root-only temp copy. Returns (rc, message)."""
import tempfile
cmd = {"sudoers": ["visudo", "-cf"], "sh": ["sh", "-n"], "bash": ["bash", "-n"], "nft": ["nft", "-c", "-f"]}[kind]
with tempfile.TemporaryDirectory(prefix="felhom-bundle-") as d:
p = os.path.join(d, "f")
with open(p, "wb") as f:
f.write(data)
r = subprocess.run(cmd + [p], capture_output=True, text=True, timeout=60)
return r.returncode, (r.stdout + r.stderr).strip()[-300:]
class Apply:
def __init__(self, runner, plan_path):
@@ -203,8 +332,12 @@ class Apply:
def check_plan(self, plan):
mode = plan.get("mode", "apply")
if mode not in ("apply", "inventory", "health", "facts", "live-restore-on"):
if mode not in ("apply", "inventory", "health", "facts", "live-restore-on", "bundle"):
raise Refused("R11", f"unknown mode {mode!r}")
if mode == "bundle":
if plan.get("layer") != "host":
raise Refused("R11", "bundle is a host-layer mode")
return mode, "host", 0, "bundle"
layer = plan.get("layer")
if layer not in ("guest", "host", "docker"):
raise Refused("R12", f"layer {layer!r} is not guest, host or docker")
@@ -298,9 +431,12 @@ class Apply:
raise Refused("R3", f"{TRUST_FILE} names no host_id")
return t
def verify_signed(self, signed, trust):
"""R3: an operator-signed os_docker_step, checked HERE (not by the agent): signature against the root-owned
signers file, op, host binding, time window, and a root-owned nonce record (no replay)."""
def verify_signed(self, signed, trust, op_name=SIGNED_OP, signers=None, burn=True):
"""R3: an operator-signed job, checked HERE (not by the agent): signature against the root-owned signers file (or,
for a bundle on a box that has none, the PINNED key — `signers` names that temp file), op, host binding, time
window, and a root-owned nonce record (no replay). burn=False leaves the nonce for the caller to burn after its
own checks (a bundle refused for a wrong sha keeps its job usable — the operator fixes the sha, not the key).
Returns the params; with burn=False, (params, nonce, expiry)."""
import base64
if not isinstance(signed, dict) or not isinstance(signed.get("blob_b64"), str) or not isinstance(signed.get("sig"), str):
raise Refused("R3", "the signed job is malformed")
@@ -312,17 +448,19 @@ class Apply:
key_id = op.get("key_id", "")
if not isinstance(key_id, str) or not re.match(r"^[A-Za-z0-9._-]{1,64}$", key_id):
raise Refused("R3", "the signed blob names no plain key_id")
try:
st = self.r.stat(TRUST_SIGNERS)
except OSError:
raise Refused("R3", f"no {TRUST_SIGNERS} — no operator key to check a signed job against")
if st.st_uid != 0 or (st.st_mode & 0o022):
raise Refused("R3", f"{TRUST_SIGNERS} is not root-owned and root-only-writable")
rc = self.r.verify_sig(TRUST_SIGNERS, key_id, SIG_NAMESPACE, blob, signed["sig"])
if signers is None:
signers = TRUST_SIGNERS
try:
st = self.r.stat(TRUST_SIGNERS)
except OSError:
raise Refused("R3", f"no {TRUST_SIGNERS} — no operator key to check a signed job against")
if st.st_uid != 0 or (st.st_mode & 0o022):
raise Refused("R3", f"{TRUST_SIGNERS} is not root-owned and root-only-writable")
rc = self.r.verify_sig(signers, key_id, SIG_NAMESPACE, blob, signed["sig"])
if rc != 0:
raise Refused("R3", f"the operator signature does not verify (ssh-keygen rc={rc})")
if op.get("op") != SIGNED_OP:
raise Refused("R3", f"the signed op is {op.get('op')!r}, not {SIGNED_OP}")
if op.get("op") != op_name:
raise Refused("R3", f"the signed op is {op.get('op')!r}, not {op_name}")
if (op.get("target") or {}).get("host_id") != trust["host_id"]:
raise Refused("R3", "the signed job is for another host")
now = self.r.now()
@@ -336,12 +474,20 @@ class Apply:
nonce = op.get("nonce")
if not isinstance(nonce, str) or not nonce:
raise Refused("R3", "the signed job has no nonce")
if nonce in self.r.read_nonces():
raise Refused("R3", "the signed job was already used (replay)")
if not burn:
return op.get("params") or {}, nonce, exp
self.burn_nonce(nonce, exp)
return op.get("params") or {}
def burn_nonce(self, nonce, exp):
seen = self.r.read_nonces()
if nonce in seen:
raise Refused("R3", "the signed job was already used (replay)")
seen[nonce] = exp
now = self.r.now()
self.r.write_nonces({k: v for k, v in seen.items() if v > now})
return op.get("params") or {}
def docker_authority(self, plan):
"""R3 for the docker layer: returns (who, undo). A signed job binds the EXACT package list and the undo flag."""
@@ -481,6 +627,10 @@ class Apply:
for k, src in (("debian", "debian"), ("docker_engine", "engine"), ("containerd", "containerd")):
g[k] = kv.get(src, "").strip() or "unknown"
g["live_restore"] = {"true": "on", "false": "off"}.get(kv.get("live", "").strip(), "unknown")
try:
h["config_bundle"] = Bundle(self).state()
except Exception as e: # a read problem must never cost the System page its other facts
h["config_bundle"] = {"version": "unknown", "error": str(e)[:200]}
self.report["facts"] = {"host": h, "guest": g}
return 0
@@ -660,6 +810,8 @@ class Apply:
self.report.update(mode=self.mode, layer=self.layer, release_id=plan.get("release_id"), vmid=self.vmid)
if self.mode == "facts":
return self.facts()
if self.mode == "bundle":
return Bundle(self).from_plan(plan)
if self.layer == "host":
self.check_appliance()
self.check_guest(self.vmid)
@@ -902,8 +1054,341 @@ class Apply:
self.x(APT_ENV + ["apt-get", "-q", "update"], timeout=600)
def main(argv, runner=None):
class Bundle:
"""The config bundle (R-840, `11` §5.4.2): every root-owned file the installer's step 5 writes, installed as ONE
signed unit. Every check runs before the first write; a failed write or a failed self-check puts every previous
copy back. Refusal codes: R1 (the bundle file), R3 (authority), R16 (a path outside BUNDLE_FILES), R17 (a trust
file), R18 (a content check or a wrong sha)."""
def __init__(self, apply):
self.a, self.r = apply, apply.r
self.report = apply.report
# ---------- reading and checking ----------
def load_file(self, path):
d, base = os.path.dirname(path or ""), os.path.basename(path or "")
if d != PLAN_DIR or not BUNDLE_RE.match(base) or ".." in path:
raise Refused("R1", f"the bundle must be {PLAN_DIR}/bundle-<version>.json, got {path!r}")
try:
st = self.r.stat(path)
except OSError as e:
raise Refused("R1", f"cannot stat the bundle: {e}")
if not stat.S_ISREG(st.st_mode):
raise Refused("R1", "the bundle is not a regular file")
if st.st_uid != self.r.agent_uid():
raise Refused("R1", f"the bundle is not owned by {AGENT_USER}")
if st.st_size > BUNDLE_MAX_BYTES:
raise Refused("R1", "the bundle is larger than 4 MB")
return self.r.read_bytes(path)
def parse(self, data, want_sha, want_version=None):
"""The bundle bytes → [(dest, content, mode, check, policy)], every content check passed. Nothing is written."""
import base64
got = sha256_hex(data)
if got != want_sha:
raise Refused("R18", f"the bundle's sha256 is {got}, not the pinned {want_sha}")
try:
b = json.loads(data)
except ValueError as e:
raise Refused("R18", f"the bundle is not JSON: {e}")
if not isinstance(b, dict) or b.get("format") != BUNDLE_FORMAT:
raise Refused("R18", f"the bundle format is not {BUNDLE_FORMAT}")
ver = b.get("agent_version")
if not isinstance(ver, str) or not re.match(r"^[0-9]+\.[0-9]+\.[0-9]+(-[0-9A-Za-z.]+)?$", ver):
raise Refused("R18", f"the bundle names no agent version ({ver!r})")
if want_version is not None and ver != want_version:
raise Refused("R18", f"the bundle is for agent {ver}, the signed job pins {want_version}")
files = b.get("files")
if not isinstance(files, list) or not files:
raise Refused("R18", "the bundle lists no files")
out, seen = [], set()
for e in files:
if not isinstance(e, dict) or not isinstance(e.get("path"), str):
raise Refused("R18", "a bundle entry has no path")
dest = e["path"]
if dest in TRUST_PATHS:
raise Refused("R17", f"{dest} is the trust root — no bundle may add, remove or change a signer")
if dest not in BUNDLE_DESTS:
raise Refused("R16", f"{dest} is not a path a bundle may write")
if dest in seen:
raise Refused("R18", f"{dest} is named twice")
seen.add(dest)
try:
content = base64.b64decode(e.get("content_b64", ""), validate=True)
except (ValueError, TypeError):
raise Refused("R18", f"{dest}: the content is not base64")
if sha256_hex(content) != e.get("sha256"):
raise Refused("R18", f"{dest}: the content does not match its sha256")
_, _, mode, check, policy = BUNDLE_DESTS[dest]
self.check_content(dest, check, content)
out.append((dest, content, mode, check, policy))
order = [x[0] for x in BUNDLE_FILES]
out.sort(key=lambda x: order.index(x[0])) # the table's order: the sudoers files last
return ver, out
def check_content(self, dest, check, content):
try:
text = content.decode("utf-8")
except UnicodeDecodeError:
raise Refused("R18", f"{dest} is not UTF-8 text")
if check in ("sudoers", "sh", "bash", "nft"):
rc, msg = self.r.check_content(check, content)
if rc != 0:
raise Refused("R18", f"{dest} fails its {check} check: {msg}")
elif check == "python":
try:
compile(text, dest, "exec")
except SyntaxError as e:
raise Refused("R18", f"{dest} is not valid Python: {e}")
elif check in ("unit", "unit-nort", "agent-unit", "dropin"):
if not re.search(r"^\[(Unit|Service|Timer)\]\s*$", text, re.M):
raise Refused("R18", f"{dest} has no [Unit]/[Service]/[Timer] section")
if check == "unit-nort" and re.search(r"^\s*RuntimeDirectory\s*=", text, re.M | re.I):
raise Refused("R18", f"{dest} declares RuntimeDirectory= — the G1 incident; refused")
if check == "agent-unit" and not re.search(r"^User=" + AGENT_USER + r"\s*$", text, re.M):
raise Refused("R18", f"{dest} does not run the agent as {AGENT_USER}")
if check == "dropin" and not re.search(r"^\[Unit\]\s*$", text, re.M):
raise Refused("R18", f"{dest} must keep its start limits in [Unit] (SF-3)")
# The route must survive the bundle: a sudoers without this wrapper's line, or a wrapper without the bundle
# mode, would cut the box off from every later bundle.
if dest == "/etc/sudoers.d/felhom-agent" and SELF_CHECK_SUDO_LINE not in text:
raise Refused("R18", "the new sudoers no longer lets the agent call felhom-os-apply — the bundle route would end")
if dest == "/usr/local/sbin/felhom-os-apply" and f'BUNDLE_OP = "{BUNDLE_OP}"' not in text:
raise Refused("R18", "the new felhom-os-apply has no bundle mode — the bundle route would end")
def live(self, dest):
"""(bytes or None, mode or None, uid or None) of what the box has now."""
if not self.r.lexists(dest):
return None, None, None
st = self.r.stat(dest)
if not stat.S_ISREG(st.st_mode):
return b"", None, None # a symlink or a directory: always replaced by the real file
return self.r.read_bytes(dest), stat.S_IMODE(st.st_mode), st.st_uid
def plan_writes(self, files):
"""Decide per file: write / same / kept (if-absent and present) / skipped (oob on a box without the belt)."""
oob = self.r.isdir(OOB_DIR)
plan = []
for dest, content, mode, check, policy in files:
if policy == "oob" and not oob:
plan.append((dest, content, mode, "skipped", None, None))
continue
old, old_mode, old_uid = self.live(dest)
if policy == "if-absent" and old is not None:
plan.append((dest, content, mode, "kept", old, old_mode))
elif old == content and old_mode == mode and old_uid == 0:
plan.append((dest, content, mode, "same", old, old_mode))
else:
plan.append((dest, content, mode, "write", old, old_mode))
return plan
# ---------- the two entries ----------
def from_plan(self, plan):
"""A signed agent_config_update, from the agent (sudo, the --plan line)."""
data = self.load_file(plan.get("bundle"))
trust = self.a.load_trust()
signers, bootstrap = TRUST_SIGNERS, False
if not self.r.lexists(TRUST_SIGNERS):
signers, bootstrap = PINNED_SIGNERS, True
self.r.log(f"os-apply: BUNDLE {TRUST_SIGNERS} is missing — verifying against the installer's pinned key "
f"{PINNED_OPERATOR_KEY_ID} only")
params, nonce, exp = self.a.verify_signed(plan.get("signed"), trust, op_name=BUNDLE_OP,
signers=None if not bootstrap else signers, burn=False)
sha, ver = params.get("bundle_sha256"), params.get("agent_version")
if not isinstance(sha, str) or not re.match(r"^[0-9a-f]{64}$", sha) or not isinstance(ver, str):
raise Refused("R3", "the signed job does not pin agent_version and bundle_sha256")
ver, files = self.parse(data, sha, ver)
self.a.burn_nonce(nonce, exp)
return self.install(ver, sha, files, "signed", bootstrap)
def from_installer(self, path, sha):
"""The installer (root, never through sudo): the hub manifest pinned the sha; no signature."""
try:
data = self.r.read_bytes(path)
except OSError as e:
raise Refused("R1", f"cannot read the bundle: {e}")
ver, files = self.parse(data, sha)
return self.install(ver, sha, files, "installer", False)
# ---------- install, self-check, undo ----------
def install(self, ver, sha, files, authority, bootstrap):
log = self.r.log
plan = self.plan_writes(files)
counts = {k: sum(1 for p in plan if p[3] == k) for k in ("write", "same", "kept", "skipped")}
log(f"os-apply: BUNDLE START agent={ver} sha={sha[:16]} authority={authority} files={len(plan)} "
f"write={counts['write']} same={counts['same']} kept={counts['kept']} skipped={counts['skipped']}")
stamp = time.strftime("%Y%m%dT%H%M%SZ", time.gmtime(self.r.now()))
prev = os.path.join(BUNDLE_PREV_DIR, f"{stamp}-before-{ver}")
done = [] # (dest, old bytes or None, old mode)
rep = {"agent_version": ver, "sha256": sha, "authority": authority, "prev_dir": prev,
"written": [p[0] for p in plan if p[3] == "write"], "kept": [p[0] for p in plan if p[3] == "kept"],
"skipped": [p[0] for p in plan if p[3] == "skipped"], "same": counts["same"]}
self.report["bundle"] = rep
try:
for dest, content, mode, action, old, old_mode in plan:
if action != "write":
continue
if old is not None:
self.r.put_file(prev + dest, old, 0o600)
done.append((dest, old, old_mode))
self.r.put_file(dest, content, mode)
log(f"os-apply: BUNDLE WROTE {dest} ({'replaced' if old is not None else 'new'})")
self.after_install(plan)
rep["self_check"] = self.self_check(plan)
except (Refused, OSError, subprocess.SubprocessError) as e:
reason = e.reason if isinstance(e, Refused) else str(e)
log(f"os-apply: BUNDLE FAILED — {reason}; putting {len(done)} previous file(s) back")
rep["rolled_back"] = self.undo(done)
rep["failed"] = reason[:300]
self.report["failed"] = {"rc": 3, "step": "bundle", "reason": reason[:300]}
return 3
rep["signers_created"] = False
if bootstrap and not self.r.lexists(TRUST_SIGNERS):
self.r.put_file(TRUST_SIGNERS, pinned_signers_line().encode(), 0o644)
rep["signers_created"] = True
log(f"os-apply: BUNDLE created {TRUST_SIGNERS} with exactly the pinned key {PINNED_OPERATOR_KEY_ID}")
record = {"format": BUNDLE_FORMAT, "agent_version": ver, "bundle_sha256": sha, "authority": authority,
"installed_at": time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime(self.r.now())),
"files": {p[0]: (sha256_hex(p[1]) if p[3] in ("write", "same") else
(sha256_hex(p[4]) if p[3] == "kept" else "skipped")) for p in plan}}
self.r.put_file(BUNDLE_RECORD, (json.dumps(record, indent=2, sort_keys=True) + "\n").encode(), 0o644)
self.prune()
log(f"os-apply: BUNDLE DONE agent={ver} written={counts['write']} same={counts['same']} "
f"self-check=ok signers-created={rep['signers_created']}")
return 0
def after_install(self, plan):
written = {p[0] for p in plan if p[3] == "write"}
dests = {p[0] for p in plan if p[3] != "skipped"}
if any(d.startswith("/etc/systemd/system/") for d in written):
self.must(["systemctl", "daemon-reload"], "systemctl daemon-reload")
if "/etc/tmpfiles.d/felhom-privsep.conf" in written:
self.must(["systemd-tmpfiles", "--create", "/etc/tmpfiles.d/felhom-privsep.conf"], "systemd-tmpfiles")
# `enable --now` starts a unit that is not running and leaves a running one alone (no restart). For the crash
# guard that runs its boot step once on a box that never had it: kernel.panic for THIS boot, as the boot unit
# would set it (`11` §5.9); with no earlier state it is the "first" boot, never counted as an unclean one.
if "/etc/systemd/system/felhom-crash-guard.service" in dests:
self.must(["systemctl", "enable", "--now", "felhom-crash-guard.service", "felhom-crash-guard-check.timer"],
"enable the crash guard")
if "/etc/systemd/system/felhom-mgmt-watchdog.timer" in dests:
self.must(["systemctl", "enable", "--now", "felhom-mgmt-watchdog.timer"], "enable the mgmt watchdog timer")
def must(self, argv, what):
rc, out, err = self.r.host(argv, 120)
if rc != 0:
raise Refused("R18", f"{what} failed (rc={rc}): {(out + err).strip()[-200:]}")
def self_check(self, plan):
"""After the install, before the record: the box still has a working route and working wrappers."""
sc = {}
self.must(["visudo", "-c"], "visudo -c (the whole sudoers)")
sc["visudo"] = "ok"
rc, out, err = self.r.host(["sudo", "-n", "-l", "-U", AGENT_USER], 60)
if rc != 0 or "/usr/local/sbin/felhom-os-apply --plan" not in out:
raise Refused("R18", f"sudo -l for {AGENT_USER} no longer lists felhom-os-apply (rc={rc})")
sc["sudo_l"] = "felhom-os-apply listed"
rc, out, err = self.r.host(["/usr/bin/python3", "/usr/local/sbin/felhom-os-apply", "--self-check"], 60)
if rc != 0 or f"bundle-format={BUNDLE_FORMAT}" not in out:
raise Refused("R18", f"the installed felhom-os-apply does not answer its self-check (rc={rc})")
sc["os_apply"] = out.strip()[:120]
rc, out, err = self.r.host(["/bin/sh", "/usr/local/sbin/felhom-selfupdate-guarded"], 60)
if rc != 2 or "usage" not in (out + err):
raise Refused("R18", f"the installed felhom-selfupdate-guarded does not answer with its usage (rc={rc})")
sc["selfupdate"] = "usage ok"
dests = {p[0] for p in plan if p[3] != "skipped"}
if "/usr/local/sbin/felhom-crash-guard" in dests:
rc, out, err = self.r.host(["/usr/local/sbin/felhom-crash-guard", "status"], 60)
try:
st = json.loads(out)
except ValueError:
st = None
if rc != 0 or not isinstance(st, dict):
raise Refused("R18", f"the crash guard does not answer its status (rc={rc})")
try:
panic = int(self.r.read_file("/proc/sys/kernel/panic").strip())
except (OSError, ValueError):
panic = None
if panic != st.get("kernel_panic"):
raise Refused("R18", f"kernel.panic is {panic}, the crash guard set {st.get('kernel_panic')}")
sc["crash_guard"] = {"armed": st.get("armed"), "kernel_panic": panic}
return sc
def undo(self, done):
"""Put every previous copy back (newest write first); remove files the bundle created."""
back = []
for dest, old, old_mode in reversed(done):
try:
if old is None:
self.r.remove(dest)
else:
self.r.put_file(dest, old, old_mode if old_mode is not None else 0o644)
back.append(dest)
except OSError as e:
self.r.log(f"os-apply: BUNDLE UNDO could not restore {dest}: {e}")
if any(d.startswith("/etc/systemd/system/") for d in back):
self.r.host(["systemctl", "daemon-reload"], 120)
rc, _, _ = self.r.host(["visudo", "-c"], 60)
self.r.log(f"os-apply: BUNDLE UNDO restored={len(back)} visudo-after={'ok' if rc == 0 else 'FAILED'}")
return back
def prune(self):
names = [n for n in self.r.list_dir(BUNDLE_PREV_DIR)]
for n in names[:-BUNDLE_KEEP]:
self.r.rmtree(os.path.join(BUNDLE_PREV_DIR, n))
# ---------- the facts mode's view ----------
def state(self):
"""What the box runs: the record, and every bundle path's live sha256 against it (drift = changed by hand)."""
rec = None
try:
rec = json.loads(self.r.read_file(BUNDLE_RECORD))
except (OSError, ValueError):
rec = None
live = {}
for dest, *_ in BUNDLE_FILES:
try:
data, _, _ = self.live(dest)
except OSError:
data = None
live[dest] = sha256_hex(data) if data is not None else "absent"
out = {"version": (rec or {}).get("agent_version", "none"), "live": live}
if rec:
out["installed_at"] = rec.get("installed_at")
out["bundle_sha256"] = rec.get("bundle_sha256")
want = rec.get("files") or {}
out["drift"] = sorted(d for d, h in want.items() if h != "skipped" and live.get(d) != h)
out["signers_present"] = self.r.lexists(TRUST_SIGNERS)
return out
def main(argv, runner=None, environ=None):
r = runner or Runner()
env = os.environ if environ is None else environ
if argv[1:] == ["--self-check"]:
# The bundle's self-check runs the NEW wrapper this way: it parses and starts. Reads nothing, changes nothing.
print(f"felhom-os-apply ok bundle-format={BUNDLE_FORMAT} files={len(BUNDLE_FILES)}")
return 0
if len(argv) == 5 and argv[1] == "--install-bundle" and argv[3] == "--sha256":
# The INSTALLER's entry (root, a fresh box). Unreachable through sudo: the sudoers line pins argv[1] to --plan,
# and a sudo caller is refused here as well (sudo always sets SUDO_UID).
a = Apply(r, "")
if "SUDO_UID" in env:
r.log("os-apply: REFUSED: R1 --install-bundle is the installer's entry, never through sudo")
print("OSAPPLY-REPORT " + json.dumps({"refused": {"code": "R1", "reason": "install-bundle via sudo"}}))
return 2
sha = argv[4]
if not re.match(r"^[0-9a-f]{64}$", sha):
print("OSAPPLY-REPORT " + json.dumps({"refused": {"code": "R18", "reason": "sha256 is not 64 lowercase hex"}}))
return 2
a.report["mode"] = "bundle"
try:
rc = Bundle(a).from_installer(argv[2], sha)
except Refused as e:
r.log(f"os-apply: REFUSED: {e.code} {e.reason}")
a.report["refused"] = {"code": e.code, "reason": e.reason}
rc = 2
print("OSAPPLY-REPORT " + json.dumps(a.report, sort_keys=True))
return rc
if len(argv) != 3 or argv[1] != "--plan":
r.log("os-apply: REFUSED: R1 usage: felhom-os-apply --plan /var/lib/felhom-agent/os/plan-<id>.json")
print("OSAPPLY-REPORT " + json.dumps({"refused": {"code": "R1", "reason": "usage"}}))