Compare commits

..

2 Commits

Author SHA1 Message Date
admin c9fa2e717b 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
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
2026-10-04 19:36:58 +02:00
admin 0af1187e03 CHANGELOG + REPORT: v0.142.1 released (R-858)
gates / gates (push) Successful in 19s
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
2026-10-04 18:21:46 +02:00
17 changed files with 1574 additions and 52 deletions
+18
View File
@@ -1,3 +1,21 @@
## v0.142.1 — a Docker step no longer leaves the controller and traefik blind (R-858, `09` decision 95)
> **RELEASED 2026-10-04** by `scripts/release-agent.sh` — tag `v0.142.1` (`4950030`), sha256
> `003f882a59abfc10021a1f97a4744ab78ab1e916b1392dcef132e7067f3c62bd`, verified by download. Not vouched at release time.
**MinAgent impact:** none. A same-day patch release, by the operator's ruling 95 after a live incident.
- **The incident.** v0.142.0's Docker step on demo-felhom (14:13 UTC) restarted dockerd, which recreates
`/run/docker.sock`. live-restore kept every container running — including `felhom-controller` and `traefik`, which
bind-mount the socket FILE and so kept the deleted inode. The controller could not reach Docker for 1 h 44 min (hub:
DOWN, "docker not reachable"); its own health check stayed "healthy", so the step's health rule PASSED.
- **The fix.** After a Docker step that installed something, the wrapper restarts ONLY the containers that mount
`/var/run/docker.sock` or `/run/docker.sock` (`os-apply: SOCKET-USERS restarted=…`; the apps and the engine untouched).
The guest health reading gains `controller_docker_ok` (`docker exec felhom-controller docker version`), and the health
rule fails when it is false — the consequence, not the mechanism.
- Proven live on demo-hp BEFORE the release (signed undo to 29.7.2): the two restarted, guest / controller / traefik on
the same socket inode, healthy. Red-proofs: `felhom.eu/documentation/audits/os-docker-crash-2026-10-04/partE-incident/`.
## v0.142.0 — the Docker engine slow lane, the version report, the crash guard (`11` §5.8, §5.9; `09` decisions 87–89)
> **RELEASED 2026-10-04** by `scripts/release-agent.sh` — tag `v0.142.0` (`b1746c2`), sha256
+2 -9
View File
@@ -1,9 +1,2 @@
# REPORT — 2026-10-04: v0.142.0, Docker slow lane + version report + crash guard
Full session report: `felhom.eu/REPORT-os-docker-crash-2026-10-04.md`.
- Docker live-restore turned on by reload (never a restart); the Docker engine set as a slow lane whose authority the
root wrapper checks itself (signed job against a root-owned key file, or the root-owned ring-0 mark); same-id health.
- The box reports its versions (Proxmox, kernels, Debian, Docker, live-restore, held packages, taint, crash guard).
- The crash guard: a crashed host restarts, the 3rd unclean stop within an hour leaves it off, 24 h re-arm.
- Tests and 17 red-proofs; live on both demo boxes (see the session report).
- v0.142.1 (same day, ruling 95): a Docker step restarts the containers that mount the docker socket, and the health
rule checks the controller reaches Docker (R-858, found live by the operator on demo-felhom).
+2
View File
@@ -15,6 +15,8 @@
| `SudoHostOps.run` | internal/storage/hostops.go | `run(ctx, name, args...) error` | allowlisted exec with stderr-wrapped error | Every arg pre-validated via validate.go before this is called |
| `Prober.Probe` | internal/capability/probe.go | `Probe(ctx) []Status` | live sudo-policy capability check (`sudo -n -l --`) | Needs a DIRECT runner (never the sudo-prefixing one — double-sudo); never executes probed cmds. v0.86.0: config-gated caps (`Capability.GatedBy` + `Prober.GateActive`) report `inactive`/"disabled by configuration" ONLY when healthy — broken plumbing stays degraded; the pbsdr-* gate answers from `pbsdr.Manager.DRConfigured` (marker-backed across restarts) |
| `stageTemp` | internal/localapi/intermediary.go | `stageTemp(pattern, content) (path, err)` | random-named temp before a root `install` (audit B1) | Fixed /tmp names are a TOCTOU — sudoers globs expect `/tmp/felhom-*-*.ext` |
| `BUNDLE_FILES` + `Bundle` (mode `bundle`, `--install-bundle`) | configs/felhom-os-apply | the ONE table of root-owned paths + the installer of them | ANY new root-owned file the installer writes (sudoers line, wrapper, unit) — add it to the table, never a new installer fetch (R-840) | The builder (`scripts/build-config-bundle.py`) and the installer read the same table; `test_every_root_file_the_installer_writes_is_in_the_bundle` fails on a path the bundle lacks. Trust files (`/etc/felhom/os-trust.json`, `operator-signers`) are NEVER bundle paths (R17) |
| `osupdate.ConfigUpdateExecutor` | internal/osupdate/bundle.go | signed op `agent_config_update` {agent_version, bundle_sha256} | delivering the bundle to an installed box | a courier only: the root wrapper re-verifies signature, host, nonce and sha itself |
| `guesthook.InstallSnippet` / `Register` | internal/guesthook/install.go | `InstallSnippet(ctx, runner) error` | pre-start self-heal hook install (C1 net) | Same random-temp+install pattern; snippet delegates to the agent binary (no shell logic). Issues `mkdir -p /var/lib/vz/snippets` FIRST (v0.63.0, B2 — fresh boxes lack the dir; sudoers grants exactly that argv) |
### Disk / format safety (role gates, durable IDs, format guards)
+12 -1
View File
@@ -1101,7 +1101,18 @@ func runDaemon(cfg config.Config, logger *slog.Logger, logRing *applog.Ring) int
}
return release, nil
}}
jobsRunner := signedjobs.NewRunner(client, gate, signedjobs.ExecutorChain{wipeExec, decommExec, updateExec, dockerExec}, cfg.Hub.HostID, logger)
// Agent v0.143.0 (R-840): the config bundle — the box's root-owned files by a signed job; the wrapper verifies it.
bundleExec := osupdate.ConfigUpdateExecutor{Leg: osLeg, URLTemplate: suCfg.URLTemplate, Username: suCfg.Username, Token: suCfg.Token,
// The capability probe confirms from the agent's side: `sudo -l` lists every command the new sudoers grants.
AfterInstall: func(ctx context.Context) {
ok, total, degraded := capability.Summarize(probeAll(ctx))
names := make([]string, 0, len(degraded))
for _, d := range degraded {
names = append(names, d.Name)
}
logger.Warn("osupdate: capability probe after the config bundle", "ok", ok, "total", total, "degraded", strings.Join(names, ","))
}}
jobsRunner := signedjobs.NewRunner(client, gate, signedjobs.ExecutorChain{wipeExec, decommExec, updateExec, dockerExec, bundleExec}, cfg.Hub.HostID, logger)
loop.SetEnvelopeObserver(hub.MultiObserver(desiredSyncer, jobsRunner))
// Controller-driven escrow ceremony (v0.88.0): static config facts + the LATE-BOUND DR gate —
+13 -1
View File
@@ -43,7 +43,7 @@ func main() {
func run() error {
var (
op = flag.String("op", "", "op class to sign, e.g. storage_wipe | guest_destroy | decommission | agent_update")
op = flag.String("op", "", "op class to sign, e.g. storage_wipe | guest_destroy | decommission | agent_update | os_docker_step | agent_config_update")
host = flag.String("host", "", "target host_id (anti-retarget — the op runs ONLY on this host)")
guest = flag.String("guest", "", "target guest_id (\"\" = host-scoped op)")
keyID = flag.String("key-id", "", "key id of the signing key (must match a pinned agent signer)")
@@ -52,6 +52,7 @@ func run() error {
fstype = flag.String("fstype", "ext4", "for storage_wipe: the filesystem to mkfs after wipe")
agentVer = flag.String("agent-version", "", "for agent_update: the target agent version (e.g. 0.70.1)")
sha256Hex = flag.String("sha256", "", "for agent_update: the pinned lowercase-hex sha256 of the target binary")
bundleSHA = flag.String("bundle-sha256", "", "for agent_config_update: the pinned sha256 of felhom-config-bundle.json (R-840)")
keyFile = flag.String("key", "", "operator signing key (ssh private key / sk- key handle) for ssh-keygen -Y sign")
ttl = flag.Duration("ttl", 30*time.Minute, "validity window from now (issued_at..expires_at)")
nonce = flag.String("nonce", "", "explicit nonce (default: a fresh 128-bit random nonce)")
@@ -95,6 +96,17 @@ func run() error {
}
pj, _ := json.Marshal(map[string]string{"version": *agentVer, "sha256": *sha256Hex})
params = string(pj)
case "agent_config_update":
// R-840: the box's root-owned files. The ROOT wrapper verifies this signature itself and refuses a bundle
// whose sha256 is not exactly this one.
if *agentVer == "" || *bundleSHA == "" {
return fmt.Errorf("agent_config_update needs -agent-version and -bundle-sha256 (the pinned bundle hash)")
}
if !isHex64(*bundleSHA) {
return fmt.Errorf("agent_config_update -bundle-sha256 must be 64 lowercase hex chars (got %d)", len(*bundleSHA))
}
pj, _ := json.Marshal(map[string]string{"agent_version": *agentVer, "bundle_sha256": *bundleSHA})
params = string(pj)
default:
params = "{}"
}
+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"}}))
+545
View File
@@ -0,0 +1,545 @@
#!/usr/bin/env python3
"""Tests for the config bundle (R-840, `11` §5.4.2): felhom-os-apply's `bundle` mode and `--install-bundle`, and
scripts/build-config-bundle.py. An in-memory host plays the files; nothing real is written or run. Each refusal has a
test; each test names the rule it pins. Red-proof: `audits/r840-config-bundle-2026-10-04/partB/redproof.txt`.
Run: python3 configs/test_felhom_config_bundle.py (also run by internal/osupdate's Go test)
"""
import sys
sys.dont_write_bytecode = True # importing the builder must not leave scripts/__pycache__ behind
import base64
import contextlib
import hashlib
import importlib.machinery
import importlib.util
import io
import json
import os
import pathlib
import re
import stat as statmod
import unittest
HERE = pathlib.Path(__file__).resolve().parent
REPO = HERE.parent
_loader = importlib.machinery.SourceFileLoader("osapply", os.environ.get("OSAPPLY_UNDER_TEST", str(HERE / "felhom-os-apply")))
_spec = importlib.util.spec_from_loader("osapply", _loader)
osapply = importlib.util.module_from_spec(_spec)
_loader.exec_module(osapply)
_bl = importlib.machinery.SourceFileLoader("bundlebuild", str(REPO / "scripts" / "build-config-bundle.py"))
_bs = importlib.util.spec_from_loader("bundlebuild", _bl)
builder = importlib.util.module_from_spec(_bs)
_bl.exec_module(builder)
PLAN = "/var/lib/felhom-agent/os/plan-b1.json"
BUNDLE = "/var/lib/felhom-agent/os/bundle-0.143.0.json"
HOST = "demo-hp-bb76ea"
INSTALLER = REPO.parent / "felhom.eu" / "scripts" / "felhom-host-install.sh"
class St:
def __init__(self, mode, uid):
self.st_mode, self.st_uid, self.st_size = mode, uid, 100
class Box:
"""An in-memory host. files: path -> bytes; modes/uids per path; dirs: a set."""
def __init__(self, bundle_bytes, signed, oob=False, signers=True):
self.files, self.modes, self.uids = {}, {}, {}
self.dirs = {osapply.OOB_DIR} if oob else set()
self.put(osapply.TRUST_FILE, json.dumps({"host_id": HOST, "ring0_slow_lane": False}).encode(), 0o644)
if signers:
self.put(osapply.TRUST_SIGNERS, b'felhom-op-1 namespaces="felhom-op-v1" ssh-ed25519 AAAA felhom-op-1\n', 0o644)
self.put("/proc/sys/kernel/panic", b"0\n", 0o644)
self.plan = {"release_id": "bundle-0.143.0", "layer": "host", "mode": "bundle", "bundle": BUNDLE, "signed": signed}
self.put(PLAN, json.dumps(self.plan).encode(), 0o600, uid=999)
self.put(BUNDLE, bundle_bytes, 0o600, uid=999)
self.sig_rc, self.nonces, self.clock = 0, {}, 1791115200.0 # 2026-10-04T12:00:00Z
self.calls, self.logs, self.writes = [], [], []
self.visudo_fail = False # the WHOLE sudoers (`visudo -c`) after install
self.sudo_l = " (root) NOPASSWD: /usr/local/sbin/felhom-os-apply --plan /var/lib/felhom-agent/os/plan-*.json\n"
self.guard = {"armed": True, "kernel_panic": 10}
def put(self, p, data, mode, uid=0):
self.files[p], self.modes[p], self.uids[p] = data, mode, uid
# Runner interface
def now(self):
return self.clock
def log(self, line):
self.logs.append(line)
def agent_uid(self):
return 999
def verify_sig(self, signers, key_id, ns, blob, sig):
self.verified = (signers, key_id, ns)
return self.sig_rc
def read_nonces(self):
return dict(self.nonces)
def write_nonces(self, d):
self.nonces = dict(d)
def read_file(self, p):
if p not in self.files:
raise OSError("no such file")
return self.files[p].decode()
def read_bytes(self, p):
if p not in self.files:
raise OSError("no such file")
return self.files[p]
def stat(self, p):
if p not in self.files:
raise OSError("no such file")
return St(statmod.S_IFREG | self.modes[p], self.uids[p])
def lexists(self, p):
return p in self.files
def isdir(self, p):
return p in self.dirs
def put_file(self, p, data, mode):
self.writes.append(p)
self.put(p, data, mode)
def remove(self, p):
self.writes.append("rm " + p)
del self.files[p]
def list_dir(self, p):
return sorted({k[len(p) + 1:].split("/")[0] for k in self.files if k.startswith(p + "/")})
def rmtree(self, p):
for k in [k for k in self.files if k.startswith(p + "/")]:
del self.files[k]
def check_content(self, kind, data):
return (1, f"{kind}: syntax error") if b"BROKEN-SYNTAX" in data else (0, "")
def host(self, argv, timeout=600, stdin=None):
self.calls.append(argv)
if argv[:2] == ["visudo", "-c"]:
return (1, "", "parse error") if self.visudo_fail else (0, "ok", "")
if argv[:2] == ["sudo", "-n"]:
return 0, self.sudo_l, ""
if argv[-1] == "--self-check":
body = self.files.get("/usr/local/sbin/felhom-os-apply", b"")
return (0, "felhom-os-apply ok bundle-format=1 files=22\n", "") if b"BUNDLE_OP" in body else (1, "", "boom")
if argv[-1] == "/usr/local/sbin/felhom-selfupdate-guarded":
return 2, "", "felhom-selfupdate-guarded: usage: ...\n"
if argv[-1] == "status" and argv[0].endswith("felhom-crash-guard"):
return 0, json.dumps(self.guard), ""
if argv[:3] == ["systemctl", "enable", "--now"] and "felhom-crash-guard.service" in argv:
self.put("/proc/sys/kernel/panic", f"{self.guard['kernel_panic']}\n".encode(), 0o644)
return 0, "", ""
def signed_job(sha, version="0.143.0", host=HOST, op="agent_config_update", nonce="b1",
issued="2026-10-04T11:50:00Z", expires="2026-10-04T12:30:00Z"):
blob = json.dumps({"expires_at": expires, "issued_at": issued, "key_id": "felhom-op-1", "nonce": nonce, "op": op,
"params": {"agent_version": version, "bundle_sha256": sha},
"target": {"guest_id": "", "host_id": host}}, sort_keys=True).encode()
return {"blob_b64": base64.b64encode(blob).decode(), "sig": "-----BEGIN SSH SIGNATURE-----\nx\n-----END SSH SIGNATURE-----\n"}
def real_bundle(version="0.143.0"):
data = builder.build(version)
return data, hashlib.sha256(data).hexdigest()
def edited_bundle(edit):
"""The real bundle, with edit(files_list) applied and every sha recomputed — a SIGNED bundle with bad content."""
b = json.loads(builder.build("0.143.0"))
edit(b["files"])
for e in b["files"]:
e["sha256"] = hashlib.sha256(base64.b64decode(e["content_b64"])).hexdigest()
data = (json.dumps(b, indent=1, sort_keys=True) + "\n").encode()
return data, hashlib.sha256(data).hexdigest()
def replace_content(files, path, fn):
for e in files:
if e["path"] == path:
e["content_b64"] = base64.b64encode(fn(base64.b64decode(e["content_b64"]))).decode()
def run(box, argv=None, environ=None):
buf = io.StringIO()
with contextlib.redirect_stdout(buf):
rc = osapply.main(argv or ["felhom-os-apply", "--plan", PLAN], runner=box, environ=environ or {})
line = [l for l in buf.getvalue().splitlines() if l.startswith("OSAPPLY-REPORT ")][-1]
return rc, json.loads(line[len("OSAPPLY-REPORT "):])
def box_files(box):
return {p: box.files[p] for p in box.files if p in osapply.BUNDLE_DESTS}
class Install(unittest.TestCase):
def test_fresh_box_gets_every_file_and_a_record(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
rc, rep = run(box)
self.assertEqual(rc, 0, rep)
b = rep["bundle"]
# every path but the four OOB ones (no belt on this box)
self.assertEqual(len(b["written"]), len(osapply.BUNDLE_FILES) - 4, b)
self.assertEqual(len(b["skipped"]), 4)
self.assertEqual(box.files["/etc/sudoers.d/felhom-agent"], (REPO / "configs" / "felhom-agent.sudoers").read_bytes())
self.assertEqual(box.modes["/etc/sudoers.d/felhom-agent"], 0o440)
rec = json.loads(box.files[osapply.BUNDLE_RECORD])
self.assertEqual((rec["agent_version"], rec["bundle_sha256"], rec["authority"]), ("0.143.0", sha, "signed"))
self.assertIn("b1", box.nonces, "the job is consumed")
self.assertEqual(b["self_check"]["crash_guard"], {"armed": True, "kernel_panic": 10})
self.assertIn(["systemctl", "daemon-reload"], box.calls)
def test_sudoers_is_written_after_every_wrapper(self):
"""Order: a referenced wrapper is in place before the sudoers line that allows it."""
data, sha = edited_bundle(lambda f: f.reverse()) # the bundle lists the sudoers FIRST; the wrapper must reorder
box = Box(data, signed_job(sha))
rc, rep = run(box)
self.assertEqual(rc, 0, rep)
w = [p for p in box.writes if p in osapply.BUNDLE_DESTS]
self.assertEqual(w[-1], "/etc/sudoers.d/felhom-agent")
self.assertLess(w.index("/usr/local/sbin/felhom-os-apply"), w.index("/etc/sudoers.d/felhom-agent"))
def test_identical_box_writes_nothing(self):
"""The demo boxes' case: hand-copied files equal to the release → 0 written, all 'same'."""
data, sha = real_bundle()
box = Box(data, signed_job(sha))
for dest, src, mode, _, policy in osapply.BUNDLE_FILES:
if policy != "oob":
box.put(dest, (REPO / "configs" / src).read_bytes(), mode)
rc, rep = run(box)
self.assertEqual(rc, 0, rep)
self.assertEqual(rep["bundle"]["written"], [])
self.assertEqual(rep["bundle"]["same"], len(osapply.BUNDLE_FILES) - 5) # 4 oob skipped + crash-guard.conf kept
self.assertEqual(rep["bundle"]["kept"], ["/etc/felhom/crash-guard.conf"])
def test_a_wrong_mode_is_rewritten(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
box.put("/etc/sudoers.d/felhom-agent", (REPO / "configs" / "felhom-agent.sudoers").read_bytes(), 0o644)
rc, rep = run(box)
self.assertIn("/etc/sudoers.d/felhom-agent", rep["bundle"]["written"])
self.assertEqual(box.modes["/etc/sudoers.d/felhom-agent"], 0o440)
def test_tuned_crash_guard_conf_is_kept(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
box.put("/etc/felhom/crash-guard.conf", b"LIMIT=5\n", 0o644)
rc, rep = run(box)
self.assertEqual(rc, 0, rep)
self.assertEqual(box.files["/etc/felhom/crash-guard.conf"], b"LIMIT=5\n")
def test_oob_files_only_on_a_box_with_the_belt(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha), oob=True)
rc, rep = run(box)
self.assertEqual(rc, 0, rep)
self.assertIn("/etc/sudoers.d/felhom-op", rep["bundle"]["written"])
self.assertEqual(rep["bundle"]["skipped"], [])
def test_previous_copies_are_kept(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
box.put("/usr/local/sbin/felhom-pbs-apply", b"#!/bin/bash\necho old\n", 0o755)
rc, rep = run(box)
self.assertEqual(rc, 0, rep)
self.assertEqual(box.files[rep["bundle"]["prev_dir"] + "/usr/local/sbin/felhom-pbs-apply"], b"#!/bin/bash\necho old\n")
class Refusals(unittest.TestCase):
"""Each: refused, and NOTHING on the box changed."""
def refused(self, box, code):
before = dict(box_files(box))
rc, rep = run(box)
self.assertEqual(rc, 2, rep)
self.assertEqual(rep["refused"]["code"], code, rep)
self.assertEqual(box_files(box), before, "a refusal changed a file")
self.assertNotIn(osapply.BUNDLE_RECORD, box.files)
return rep
def test_wrong_sha_is_refused(self):
data, sha = real_bundle()
box = Box(data, signed_job("0" * 64))
self.refused(box, "R18")
self.assertEqual(box.nonces, {}, "a wrong sha must not burn the job")
def test_bad_signature_is_refused(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
box.sig_rc = 255
self.refused(box, "R3")
def test_other_op_is_refused(self):
data, sha = real_bundle()
self.refused(Box(data, signed_job(sha, op="agent_update")), "R3")
def test_other_host_is_refused(self):
data, sha = real_bundle()
self.refused(Box(data, signed_job(sha, host="demo-felhom-8363b5")), "R3")
def test_expired_job_is_refused(self):
data, sha = real_bundle()
self.refused(Box(data, signed_job(sha, expires="2026-10-04T11:55:00Z")), "R3")
def test_replayed_job_is_refused(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
box.nonces = {"b1": box.clock + 600}
self.refused(box, "R3")
def test_version_mismatch_is_refused(self):
data, sha = real_bundle()
self.refused(Box(data, signed_job(sha, version="0.142.1")), "R18")
def test_sudoers_failing_visudo_is_refused(self):
data, sha = edited_bundle(lambda f: replace_content(f, "/etc/sudoers.d/felhom-agent", lambda c: c + b"BROKEN-SYNTAX\n"))
self.refused(Box(data, signed_job(sha)), "R18")
def test_sudoers_dropping_the_route_is_refused(self):
data, sha = edited_bundle(lambda f: replace_content(f, "/etc/sudoers.d/felhom-agent",
lambda c: c.replace(b"/usr/local/sbin/felhom-os-apply --plan", b"/bin/true --plan")))
self.refused(Box(data, signed_job(sha)), "R18")
def test_wrapper_without_bundle_mode_is_refused(self):
data, sha = edited_bundle(lambda f: replace_content(f, "/usr/local/sbin/felhom-os-apply",
lambda c: c.replace(b'BUNDLE_OP = "agent_config_update"', b'BUNDLE_OPX = 1')))
self.refused(Box(data, signed_job(sha)), "R18")
def test_python_syntax_error_is_refused(self):
data, sha = edited_bundle(lambda f: replace_content(f, "/usr/local/sbin/felhom-crash-guard", lambda c: c + b"\ndef (\n"))
self.refused(Box(data, signed_job(sha)), "R18")
def test_unit_with_runtime_directory_is_refused(self):
data, sha = edited_bundle(lambda f: replace_content(f, "/etc/systemd/system/felhom-mgmt-watchdog.service",
lambda c: c + b"RuntimeDirectory=sshd\n"))
self.refused(Box(data, signed_job(sha)), "R18")
def test_agent_unit_not_as_the_agent_user_is_refused(self):
data, sha = edited_bundle(lambda f: replace_content(f, "/etc/systemd/system/felhom-agent.service",
lambda c: c.replace(b"User=felhom-agent", b"User=root")))
self.refused(Box(data, signed_job(sha)), "R18")
def test_a_bundle_that_changes_a_signer_is_refused(self):
"""R17: the trust root is not a bundle's to change — not even a signed one."""
def add(f):
f.append({"path": osapply.TRUST_SIGNERS, "content_b64": base64.b64encode(b"evil-key\n").decode()})
data, sha = edited_bundle(add)
box = Box(data, signed_job(sha))
self.refused(box, "R17")
self.assertIn(b"felhom-op-1", box.files[osapply.TRUST_SIGNERS])
def test_a_path_outside_the_table_is_refused(self):
def add(f):
f.append({"path": "/etc/shadow", "content_b64": base64.b64encode(b"root::0:0\n").decode()})
data, sha = edited_bundle(add)
self.refused(Box(data, signed_job(sha)), "R16")
def test_content_not_matching_its_sha_is_refused(self):
b = json.loads(builder.build("0.143.0"))
b["files"][0]["content_b64"] = base64.b64encode(b"#!/bin/bash\nexit 0\n").decode()
data = json.dumps(b).encode()
self.refused(Box(data, signed_job(hashlib.sha256(data).hexdigest())), "R18")
def test_bundle_outside_the_plan_dir_is_refused(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
box.plan["bundle"] = "/tmp/bundle-0.143.0.json"
box.files[PLAN] = json.dumps(box.plan).encode()
self.refused(box, "R1")
def test_bundle_not_owned_by_the_agent_is_refused(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
box.uids[BUNDLE] = 0
self.refused(box, "R1")
def test_no_trust_file_is_refused(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
del box.files[osapply.TRUST_FILE]
self.refused(box, "R3")
class SelfCheckUndo(unittest.TestCase):
def assert_restored(self, box, before, rep):
self.assertTrue(rep["bundle"]["rolled_back"], rep)
self.assertEqual(box_files(box), before, "the previous files must be back, byte for byte")
self.assertNotIn(osapply.BUNDLE_RECORD, box.files)
def with_old_files(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
box.put("/usr/local/sbin/felhom-pbs-apply", b"#!/bin/bash\necho old\n", 0o755)
box.put("/etc/sudoers.d/felhom-agent", b"# old sudoers\n", 0o440)
return box
def test_route_missing_after_install_puts_everything_back(self):
box = self.with_old_files()
before = dict(box_files(box))
box.sudo_l = " (root) NOPASSWD: /bin/true\n"
rc, rep = run(box)
self.assertEqual(rc, 3, rep)
self.assert_restored(box, before, rep)
self.assertEqual(box.calls[-1], ["visudo", "-c"], "the undo re-checks the whole sudoers")
def test_visudo_failing_after_install_puts_everything_back(self):
box = self.with_old_files()
before = dict(box_files(box))
box.visudo_fail = True
rc, rep = run(box)
self.assertEqual(rc, 3, rep)
self.assert_restored(box, before, rep)
def test_crash_guard_disagreeing_with_kernel_panic_puts_everything_back(self):
box = self.with_old_files()
before = dict(box_files(box))
box.guard = {"armed": True, "kernel_panic": 10}
box.host_orig = box.host
def host(argv, timeout=600, stdin=None):
if argv[:3] == ["systemctl", "enable", "--now"]:
box.calls.append(argv)
return 0, "", "" # the unit "started" but kernel.panic stayed 0
return box.host_orig(argv, timeout, stdin)
box.host = host
rc, rep = run(box)
self.assertEqual(rc, 3, rep)
self.assert_restored(box, before, rep)
class TrustBootstrap(unittest.TestCase):
def test_missing_signers_verifies_against_the_pinned_key_and_creates_it(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha), signers=False)
rc, rep = run(box)
self.assertEqual(rc, 0, rep)
self.assertEqual(box.verified[0], osapply.PINNED_SIGNERS, "verified against the pinned key, nothing else")
self.assertTrue(rep["bundle"]["signers_created"])
self.assertEqual(box.files[osapply.TRUST_SIGNERS], osapply.pinned_signers_line().encode())
self.assertEqual(box.modes[osapply.TRUST_SIGNERS], 0o644)
def test_missing_signers_and_a_job_the_pinned_key_did_not_sign(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha), signers=False)
box.sig_rc = 255
rc, rep = run(box)
self.assertEqual((rc, rep["refused"]["code"]), (2, "R3"))
self.assertNotIn(osapply.TRUST_SIGNERS, box.files)
def test_present_signers_are_never_touched(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
box.put(osapply.TRUST_SIGNERS, b'felhom-op-2 namespaces="felhom-op-v1" ssh-ed25519 BBBB felhom-op-2\n', 0o644)
rc, rep = run(box)
self.assertEqual(rc, 0, rep)
self.assertEqual(box.verified[0], osapply.TRUST_SIGNERS)
self.assertFalse(rep["bundle"]["signers_created"])
self.assertIn(b"felhom-op-2", box.files[osapply.TRUST_SIGNERS])
def test_agent_writable_signers_are_refused(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
box.uids[osapply.TRUST_SIGNERS] = 999
rc, rep = run(box)
self.assertEqual((rc, rep["refused"]["code"]), (2, "R3"))
def test_pinned_operator_key_equals_the_installers(self):
"""The bootstrap key is exactly the one felhom-host-install.sh pins (OPERATOR_KEY_OPERATIONAL_*)."""
text = INSTALLER.read_text()
kid = re.search(r'^OPERATOR_KEY_OPERATIONAL_ID="([^"]+)"', text, re.M).group(1)
line = re.search(r'^OPERATOR_KEY_OPERATIONAL_LINE="([^"]+)"', text, re.M).group(1)
self.assertEqual((osapply.PINNED_OPERATOR_KEY_ID, osapply.PINNED_OPERATOR_KEY_LINE), (kid, line))
# and the file format is the installer's printf, byte for byte
self.assertIn("printf '%s namespaces=\"felhom-op-v1\" %s\\n'", text)
class InstallerEntry(unittest.TestCase):
def test_installer_installs_without_a_signature(self):
data, sha = real_bundle()
box = Box(data, None)
box.put("/root/bundle.json", data, 0o600)
rc, rep = run(box, ["felhom-os-apply", "--install-bundle", "/root/bundle.json", "--sha256", sha])
self.assertEqual(rc, 0, rep)
self.assertEqual(json.loads(box.files[osapply.BUNDLE_RECORD])["authority"], "installer")
def test_installer_entry_checks_the_sha(self):
data, sha = real_bundle()
box = Box(data, None)
box.put("/root/bundle.json", data, 0o600)
rc, rep = run(box, ["felhom-os-apply", "--install-bundle", "/root/bundle.json", "--sha256", "1" * 64])
self.assertEqual((rc, rep["refused"]["code"]), (2, "R18"))
def test_installer_entry_is_refused_through_sudo(self):
data, sha = real_bundle()
box = Box(data, None)
box.put("/root/bundle.json", data, 0o600)
rc, rep = run(box, ["felhom-os-apply", "--install-bundle", "/root/bundle.json", "--sha256", sha], {"SUDO_UID": "999"})
self.assertEqual((rc, rep["refused"]["code"]), (2, "R1"))
self.assertNotIn(osapply.BUNDLE_RECORD, box.files)
def test_self_check_answers(self):
buf = io.StringIO()
with contextlib.redirect_stdout(buf):
rc = osapply.main(["felhom-os-apply", "--self-check"], runner=Box(b"", None))
self.assertEqual(rc, 0)
self.assertIn("bundle-format=1", buf.getvalue())
class Facts(unittest.TestCase):
def test_state_reports_drift_against_the_record(self):
data, sha = real_bundle()
box = Box(data, signed_job(sha))
run(box)
box.put("/usr/local/sbin/felhom-pbs-apply", b"#!/bin/bash\necho by hand\n", 0o755)
a = osapply.Apply(box, PLAN)
st = osapply.Bundle(a).state()
self.assertEqual(st["version"], "0.143.0")
self.assertEqual(st["drift"], ["/usr/local/sbin/felhom-pbs-apply"])
self.assertTrue(st["signers_present"])
def test_state_without_a_record_says_none(self):
box = Box(b"", None)
st = osapply.Bundle(osapply.Apply(box, PLAN)).state()
self.assertEqual(st["version"], "none")
self.assertNotIn("drift", st)
self.assertEqual(st["live"]["/etc/sudoers.d/felhom-agent"], "absent")
class Builder(unittest.TestCase):
def test_reproducible(self):
self.assertEqual(builder.build("0.143.0"), builder.build("0.143.0"))
def test_every_source_exists_and_every_dest_is_unique(self):
dests = [e[0] for e in osapply.BUNDLE_FILES]
self.assertEqual(len(dests), len(set(dests)))
for _, src, *_ in osapply.BUNDLE_FILES:
self.assertTrue((REPO / "configs" / src).is_file(), src)
def test_every_root_file_the_installer_writes_is_in_the_bundle(self):
"""One source of truth: a felhom root-owned path the installer names must be a bundle path, a trust file, or a
path the AGENT itself writes at run time (named here, with why). Scope is a regex over the WHOLE installer."""
text = INSTALLER.read_text()
found = set(re.findall(r"(/usr/local/sbin/felhom-[a-z-]+|/etc/systemd/system/felhom-[a-z.-]+|"
r"/etc/sudoers\.d/felhom-[a-z-]+|/etc/felhom-oob\.nft|/etc/tmpfiles\.d/felhom-[a-z.-]+|"
r"/etc/felhom/[a-z.-]+)", text))
agent_writes = {"/usr/local/sbin/felhom-shared-parent", "/etc/systemd/system/felhom-shared-parent.service"}
trust = {osapply.TRUST_FILE, osapply.TRUST_SIGNERS, osapply.TRUST_SIGNERS + ".tmp", osapply.BUNDLE_RECORD}
missing = sorted(p for p in found if p not in osapply.BUNDLE_DESTS and p not in agent_writes | trust)
self.assertEqual(missing, [], "the installer writes these root files, but the bundle does not carry them")
# the limits drop-in is named through $AGENT_UNIT in the installer
self.assertIn("/etc/systemd/system/felhom-agent.service.d/felhom-agent-limits.conf", osapply.BUNDLE_DESTS)
if __name__ == "__main__":
unittest.main()
+25
View File
@@ -0,0 +1,25 @@
package hub
import (
"os"
"path/filepath"
"testing"
)
// R-840: the agent reports the bundle record itself — "none" when no bundle ever reached the box (so an old wrapper
// cannot hide that), "unknown" when it cannot be read, else the version and sha the root wrapper recorded.
func TestReadBundleRecord(t *testing.T) {
dir := t.TempDir()
p := filepath.Join(dir, "config-bundle.json")
if got := string(readBundleRecord(p)); got != `{"version":"none"}` {
t.Fatalf("absent: %s", got)
}
os.WriteFile(p, []byte("{broken"), 0o644)
if got := string(readBundleRecord(p)); got != `{"version":"unknown"}` {
t.Fatalf("broken: %s", got)
}
os.WriteFile(p, []byte(`{"format":1,"agent_version":"0.143.0","bundle_sha256":"abc","installed_at":"2026-10-04T20:00:00Z","authority":"signed","files":{"/x":"y"}}`), 0o644)
if got := string(readBundleRecord(p)); got != `{"authority":"signed","bundle_sha256":"abc","installed_at":"2026-10-04T20:00:00Z","version":"0.143.0"}` {
t.Fatalf("record: %s", got)
}
}
+53 -24
View File
@@ -92,30 +92,31 @@ type GuestNetReporter interface {
// Collector builds a HostReport from read-only sources. All deps are behind narrow
// interfaces for unit testing.
type Collector struct {
px proxmoxReader
cf CloudflaredProber
storage StorageObserver
backups BackupReporter
restoreTests RestoreTestReporter
provenTests ProvenRestoreTestReporter
pbs PBSReporter
temp TempReader // slice 9: host CPU/chassis temp (nil-safe → nil temp)
capProbe func(ctx context.Context) []capability.Status // v0.44.0: privileged-capability self-check (nil → empty)
leafFP string // v0.48.0: served local-API leaf fp (static per process; "" when local API disabled)
addrEnum AddressEnumerator // v0.119.0: host interface enumeration; nil => the REAL one (see collectAddresses)
wg WireguardReporter // S3: offsite-tunnel status (nil → stanza omitted)
pbsdr PBSDRReporter // slice 2: PBS DR tier bridge state (nil → stanza omitted)
ctrlSup ControllerSupervisorReporter // R-523: in-guest controller supervisor (nil → stanza omitted)
guestNet GuestNetReporter // R-54: per-guest network watchdog (nil → stanza omitted)
selfUpdate SelfUpdateReporter // D1: agent self-update pending status (nil → false)
mgmtPlane MgmtPlaneReporter // G1: management-plane health (nil → stanza omitted)
oob OOBReporter // H1: operator-access health (nil → stanza omitted)
system SystemReporter // R-852: the box versions (nil → API fields only)
backupTarget func() ConfiguredBackupTarget // R-109: primary backup tier id (nil → recipe records unknown)
hostID string
agentVersion string
logger *slog.Logger
now func() time.Time
px proxmoxReader
cf CloudflaredProber
storage StorageObserver
backups BackupReporter
restoreTests RestoreTestReporter
provenTests ProvenRestoreTestReporter
pbs PBSReporter
temp TempReader // slice 9: host CPU/chassis temp (nil-safe → nil temp)
capProbe func(ctx context.Context) []capability.Status // v0.44.0: privileged-capability self-check (nil → empty)
leafFP string // v0.48.0: served local-API leaf fp (static per process; "" when local API disabled)
addrEnum AddressEnumerator // v0.119.0: host interface enumeration; nil => the REAL one (see collectAddresses)
wg WireguardReporter // S3: offsite-tunnel status (nil → stanza omitted)
pbsdr PBSDRReporter // slice 2: PBS DR tier bridge state (nil → stanza omitted)
ctrlSup ControllerSupervisorReporter // R-523: in-guest controller supervisor (nil → stanza omitted)
guestNet GuestNetReporter // R-54: per-guest network watchdog (nil → stanza omitted)
selfUpdate SelfUpdateReporter // D1: agent self-update pending status (nil → false)
mgmtPlane MgmtPlaneReporter // G1: management-plane health (nil → stanza omitted)
oob OOBReporter // H1: operator-access health (nil → stanza omitted)
system SystemReporter // R-852: the box versions (nil → API fields only)
bundleRecordPath string // R-840: test seam; "" = BundleRecordPath
backupTarget func() ConfiguredBackupTarget // R-109: primary backup tier id (nil → recipe records unknown)
hostID string
agentVersion string
logger *slog.Logger
now func() time.Time
}
// NewCollector builds a collector. hostID echoes config.Hub.HostID; agentVersion is
@@ -262,6 +263,33 @@ func (c *Collector) SetSystemReporter(r SystemReporter) *Collector {
return c
}
// BundleRecordPath is the root-owned record felhom-os-apply writes after a config bundle installs (R-840).
const BundleRecordPath = "/etc/felhom/config-bundle.json"
// readBundleRecord returns the record's summary: version/sha/installed_at, "none" when the file is absent, "unknown"
// when it cannot be read or parsed (never a guess).
func readBundleRecord(path string) json.RawMessage {
if path == "" {
path = BundleRecordPath
}
b, err := os.ReadFile(path)
if os.IsNotExist(err) {
return json.RawMessage(`{"version":"none"}`)
}
var rec struct {
AgentVersion string `json:"agent_version"`
BundleSHA256 string `json:"bundle_sha256"`
InstalledAt string `json:"installed_at"`
Authority string `json:"authority"`
}
if err != nil || json.Unmarshal(b, &rec) != nil || rec.AgentVersion == "" {
return json.RawMessage(`{"version":"unknown"}`)
}
out, _ := json.Marshal(map[string]string{"version": rec.AgentVersion, "bundle_sha256": rec.BundleSHA256,
"installed_at": rec.InstalledAt, "authority": rec.Authority})
return out
}
func unknownIfEmpty(s string) string {
if strings.TrimSpace(s) == "" {
return "unknown"
@@ -273,6 +301,7 @@ func unknownIfEmpty(s string) string {
func (c *Collector) systemInfo(ctx context.Context, ns proxmox.NodeStatus) *SystemInfo {
si := &SystemInfo{PVEVersion: unknownIfEmpty(ns.PVEVersion), KernelVersion: unknownIfEmpty(ns.KVersion),
ReadAt: c.now().Format(time.RFC3339)}
si.ConfigBundle = readBundleRecord(c.bundleRecordPath)
if c.system == nil {
si.FactsError = "no facts reader wired"
return si
+4
View File
@@ -255,6 +255,10 @@ type SystemInfo struct {
Facts json.RawMessage `json:"facts,omitempty"`
FactsError string `json:"facts_error,omitempty"`
ReadAt string `json:"read_at"`
// ConfigBundle is the box's root-owned config bundle record (R-840, agent v0.143.0), read by the agent itself from
// /etc/felhom/config-bundle.json (0644): {"version":"none"} on a box no bundle reached, so an OLD wrapper cannot
// hide it. The wrapper's facts carry the same record plus the drift (files changed by hand).
ConfigBundle json.RawMessage `json:"config_bundle,omitempty"`
}
// HostMetrics is the host block, sourced from proxmox NodeStatus.
+157
View File
@@ -0,0 +1,157 @@
package osupdate
import (
"context"
"crypto/sha256"
"encoding/base64"
"encoding/hex"
"encoding/json"
"fmt"
"io"
"net/http"
"os"
"path/filepath"
"regexp"
"strings"
"time"
"gitea.dooplex.hu/admin/felhom-agent/internal/signedjobs"
)
// OpConfigUpdate is the signed op that brings a box's ROOT-OWNED files (sudoers, wrappers, units) to a release's config
// bundle (R-840, `11` §5.4.2). The params pin the agent version and the bundle's sha256; the root wrapper re-verifies
// the signature, the host binding, the nonce and the sha ITSELF — this executor is only the courier.
const OpConfigUpdate = "agent_config_update"
// BundleFileName is the bundle's name beside the binary in the Gitea generic package felhom-agent/<version>/.
const BundleFileName = "felhom-config-bundle.json"
var (
bundleVersionRe = regexp.MustCompile(`^[0-9]+\.[0-9]+\.[0-9]+(-[0-9A-Za-z.]+)?$`)
bundleSHARe = regexp.MustCompile(`^[0-9a-f]{64}$`)
)
// ConfigUpdateParams are the signed params (the wrapper reads the same two names out of the signed blob).
type ConfigUpdateParams struct {
AgentVersion string `json:"agent_version"`
BundleSHA256 string `json:"bundle_sha256"`
}
// BundleURL derives the bundle's URL from the agent binary's URL template (".../felhom-agent/{version}/felhom-agent").
func BundleURL(binaryTemplate, version string) (string, error) {
if !strings.HasSuffix(binaryTemplate, "/felhom-agent") || !strings.Contains(binaryTemplate, "{version}") {
return "", fmt.Errorf("cannot derive the bundle URL from %q (want …/{version}/felhom-agent)", binaryTemplate)
}
t := strings.TrimSuffix(binaryTemplate, "felhom-agent") + BundleFileName
return strings.ReplaceAll(t, "{version}", version), nil
}
// ConfigUpdateExecutor runs a verified agent_config_update (signedjobs.Executor).
type ConfigUpdateExecutor struct {
Leg *Leg
URLTemplate string // the agent binary's template (config.SelfUpdate.URLTemplate)
Username string
Token string
HTTPClient *http.Client
// AfterInstall runs after a bundle installed (the capability re-probe; nil = none).
AfterInstall func(ctx context.Context)
}
// Execute implements signedjobs.Executor.
func (e ConfigUpdateExecutor) Execute(ctx context.Context, op string, params json.RawMessage) error {
if op != OpConfigUpdate {
return signedjobs.ErrNoExecutor
}
so, ok := signedjobs.SignedOpFrom(ctx)
if !ok {
return fmt.Errorf("agent_config_update: no signed envelope in the context — the wrapper could not verify it")
}
var p ConfigUpdateParams
if err := json.Unmarshal(params, &p); err != nil {
return fmt.Errorf("agent_config_update: bad params: %w", err)
}
if !bundleVersionRe.MatchString(p.AgentVersion) || !bundleSHARe.MatchString(p.BundleSHA256) {
return fmt.Errorf("agent_config_update: params must pin agent_version (semver) and bundle_sha256 (64 hex)")
}
lg := e.Leg.log().With("op", OpConfigUpdate, "agent_version", p.AgentVersion)
url, err := BundleURL(e.URLTemplate, p.AgentVersion)
if err != nil {
return fmt.Errorf("agent_config_update: %w", err)
}
dir := e.Leg.PlanDir
if dir == "" {
dir = DefaultPlanDir
}
if err := os.MkdirAll(dir, 0o700); err != nil {
return fmt.Errorf("agent_config_update: plan dir: %w", err)
}
path := filepath.Join(dir, "bundle-"+p.AgentVersion+".json")
start := time.Now()
got, err := e.download(ctx, url, path)
if err != nil {
_ = os.Remove(path)
return fmt.Errorf("agent_config_update: download %s: %w", url, err)
}
defer os.Remove(path)
// The agent's own check is a courtesy (an early, clear error); the wrapper's is the gate.
if got != p.BundleSHA256 {
return fmt.Errorf("agent_config_update: the downloaded bundle's sha256 is %s, the signed job pins %s — nothing installed", got, p.BundleSHA256)
}
lg.Info("osupdate: config bundle downloaded; handing it to the root wrapper", "sha256", got[:16], "duration_ms", time.Since(start).Milliseconds())
runID := "bundle-" + e.Leg.now().UTC().Format("20060102T150405Z")
wr, err := e.Leg.call(ctx, runID, map[string]any{"release_id": "bundle-" + p.AgentVersion, "layer": LayerHost,
"mode": "bundle", "bundle": path,
"signed": map[string]string{"blob_b64": base64.StdEncoding.EncodeToString(so.Blob), "sig": string(so.Sig)}})
if err != nil {
return fmt.Errorf("agent_config_update: %w", err)
}
if wr.refused() {
lg.Warn("osupdate: config bundle REFUSED by the wrapper — nothing changed", "refused", string(wr.Refused))
return fmt.Errorf("agent_config_update: refused: %s", wr.Refused)
}
if wr.failed() {
lg.Error("osupdate: config bundle FAILED — the wrapper put the previous files back", "failed", string(wr.Failed), "bundle", string(wr.Bundle))
return fmt.Errorf("agent_config_update: failed (previous files restored): %s", wr.Failed)
}
lg.Warn("osupdate: config bundle INSTALLED", "bundle", string(wr.Bundle), "pass_seconds", wr.PassSeconds)
if e.AfterInstall != nil {
e.AfterInstall(ctx)
}
return nil
}
func (e ConfigUpdateExecutor) download(ctx context.Context, url, dest string) (string, error) {
hc := e.HTTPClient
if hc == nil {
hc = &http.Client{Timeout: 2 * time.Minute}
}
req, err := http.NewRequestWithContext(ctx, http.MethodGet, url, nil)
if err != nil {
return "", err
}
if e.Username != "" || e.Token != "" {
req.SetBasicAuth(e.Username, e.Token)
}
resp, err := hc.Do(req)
if err != nil {
return "", err
}
defer resp.Body.Close()
if resp.StatusCode < 200 || resp.StatusCode >= 300 {
return "", fmt.Errorf("HTTP %d", resp.StatusCode)
}
f, err := os.OpenFile(dest, os.O_CREATE|os.O_TRUNC|os.O_WRONLY, 0o600)
if err != nil {
return "", err
}
h := sha256.New()
// 4 MB is the wrapper's own limit; read one byte more so an oversized bundle fails there, visibly.
if _, err := io.Copy(io.MultiWriter(f, h), io.LimitReader(resp.Body, 4*1024*1024+1)); err != nil {
f.Close()
return "", err
}
if err := f.Close(); err != nil {
return "", err
}
return hex.EncodeToString(h.Sum(nil)), nil
}
+155
View File
@@ -0,0 +1,155 @@
package osupdate
import (
"context"
"crypto/sha256"
"encoding/base64"
"encoding/hex"
"encoding/json"
"errors"
"io"
"net/http"
"net/http/httptest"
"os"
"strings"
"testing"
"gitea.dooplex.hu/admin/felhom-agent/internal/reconcile"
"gitea.dooplex.hu/admin/felhom-agent/internal/signedjobs"
)
// bundleWrapper plays felhom-os-apply for mode "bundle": it records the plan and the bundle file's bytes AT CALL TIME
// (the executor deletes the file afterwards), and answers with rep.
type bundleWrapper struct {
t *testing.T
rep string
plans []map[string]any
bodies [][]byte
}
func (b *bundleWrapper) Run(_ context.Context, name string, args ...string) ([]byte, []byte, error) {
if name != WrapperPath || len(args) != 2 || args[0] != "--plan" {
b.t.Fatalf("unexpected command %s %v", name, args)
}
raw, err := os.ReadFile(args[1])
if err != nil {
b.t.Fatal(err)
}
var plan map[string]any
_ = json.Unmarshal(raw, &plan)
b.plans = append(b.plans, plan)
body, _ := os.ReadFile(plan["bundle"].(string))
b.bodies = append(b.bodies, body)
return []byte("OSAPPLY-REPORT " + b.rep + "\n"), nil, nil
}
func serveBundle(t *testing.T, body []byte) (*httptest.Server, string) {
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
if r.URL.Path != "/felhom-agent/0.143.0/"+BundleFileName {
http.NotFound(w, r)
return
}
_, _ = w.Write(body)
}))
t.Cleanup(srv.Close)
sum := sha256.Sum256(body)
return srv, hex.EncodeToString(sum[:])
}
func bundleExec(t *testing.T, w *bundleWrapper, srvURL string) (ConfigUpdateExecutor, *bool) {
l, _ := newLeg(t, &fakeWrapper{t: t}, nil)
l.Runner = w
called := false
return ConfigUpdateExecutor{Leg: l, URLTemplate: srvURL + "/felhom-agent/{version}/felhom-agent",
AfterInstall: func(context.Context) { called = true }}, &called
}
func signedCtx() context.Context {
return signedjobs.WithSignedOp(context.Background(), &reconcile.SignedOp{Blob: []byte(`{"op":"agent_config_update"}`), Sig: []byte("SIG")})
}
func params(v, sha string) json.RawMessage {
p, _ := json.Marshal(ConfigUpdateParams{AgentVersion: v, BundleSHA256: sha})
return p
}
// The courier hands the wrapper the bundle bytes it downloaded and the RAW signed envelope (the wrapper verifies both
// itself), then runs the capability probe. Red-proof: drop "signed" from the plan → the plan check below fails.
func TestConfigUpdate_PassesBundleAndEnvelopeToTheWrapper(t *testing.T) {
body := []byte(`{"format":1,"agent_version":"0.143.0","files":[]}`)
srv, sha := serveBundle(t, body)
w := &bundleWrapper{t: t, rep: `{"mode":"bundle","bundle":{"agent_version":"0.143.0","written":["/etc/sudoers.d/felhom-agent"]}}`}
e, called := bundleExec(t, w, srv.URL)
if err := e.Execute(signedCtx(), OpConfigUpdate, params("0.143.0", sha)); err != nil {
t.Fatal(err)
}
p := w.plans[0]
sg, _ := p["signed"].(map[string]any)
if p["mode"] != "bundle" || p["layer"] != "host" || sg == nil || sg["sig"] != "SIG" ||
sg["blob_b64"] != base64.StdEncoding.EncodeToString([]byte(`{"op":"agent_config_update"}`)) {
t.Fatalf("plan = %v", p)
}
if string(w.bodies[0]) != string(body) || !strings.HasSuffix(p["bundle"].(string), "/bundle-0.143.0.json") {
t.Fatalf("the wrapper got %q at %v", w.bodies[0], p["bundle"])
}
if !*called {
t.Fatal("the capability probe must run after an install")
}
if _, err := os.Stat(p["bundle"].(string)); !os.IsNotExist(err) {
t.Fatal("the downloaded bundle must be removed after the call")
}
}
func TestConfigUpdate_WrongShaNeverReachesTheWrapper(t *testing.T) {
srv, _ := serveBundle(t, []byte("tampered"))
w := &bundleWrapper{t: t}
e, called := bundleExec(t, w, srv.URL)
err := e.Execute(signedCtx(), OpConfigUpdate, params("0.143.0", strings.Repeat("a", 64)))
if err == nil || len(w.plans) != 0 || *called {
t.Fatalf("err=%v plans=%d", err, len(w.plans))
}
}
func TestConfigUpdate_RefusedAndFailedAreErrors(t *testing.T) {
for _, rep := range []string{`{"refused":{"code":"R17","reason":"trust root"}}`, `{"failed":{"rc":3},"bundle":{"rolled_back":["/x"]}}`} {
srv, sha := serveBundle(t, []byte("{}"))
w := &bundleWrapper{t: t, rep: rep}
e, called := bundleExec(t, w, srv.URL)
if err := e.Execute(signedCtx(), OpConfigUpdate, params("0.143.0", sha)); err == nil || *called {
t.Fatalf("%s: err=%v called=%v", rep, err, *called)
}
}
}
func TestConfigUpdate_GuardsBeforeAnyDownload(t *testing.T) {
w := &bundleWrapper{t: t}
e, _ := bundleExec(t, w, "http://127.0.0.1:1")
if err := e.Execute(signedCtx(), "agent_update", nil); !errors.Is(err, signedjobs.ErrNoExecutor) {
t.Fatalf("another op must pass through the chain: %v", err)
}
if err := e.Execute(context.Background(), OpConfigUpdate, params("0.143.0", strings.Repeat("a", 64))); err == nil {
t.Fatal("no envelope must refuse")
}
for _, p := range []json.RawMessage{params("0.143", strings.Repeat("a", 64)), params("0.143.0", "ABC"), json.RawMessage(`nope`)} {
if err := e.Execute(signedCtx(), OpConfigUpdate, p); err == nil {
t.Fatalf("bad params %s must refuse", p)
}
}
if len(w.plans) != 0 {
t.Fatal("no wrapper call on a refusal")
}
}
func TestBundleURL(t *testing.T) {
u, err := BundleURL("https://gitea.dooplex.hu/api/packages/admin/generic/felhom-agent/{version}/felhom-agent", "0.143.0")
if err != nil || u != "https://gitea.dooplex.hu/api/packages/admin/generic/felhom-agent/0.143.0/felhom-config-bundle.json" {
t.Fatalf("%s %v", u, err)
}
if _, err := BundleURL("https://example/felhom-agent-{version}.bin", "0.143.0"); err == nil {
t.Fatal("an underivable template must refuse")
}
}
func (b *bundleWrapper) RunStdin(ctx context.Context, _ io.Reader, name string, args ...string) ([]byte, []byte, error) {
return b.Run(ctx, name, args...)
}
+1
View File
@@ -105,6 +105,7 @@ type WrapperReport struct {
Undo bool `json:"undo"`
LiveRestore json.RawMessage `json:"live_restore"`
Facts json.RawMessage `json:"facts"`
Bundle json.RawMessage `json:"bundle"` // the config bundle's result (R-840, mode "bundle")
}
func (w WrapperReport) refused() bool { return len(w.Refused) > 0 && string(w.Refused) != "null" }
+1 -1
View File
@@ -346,7 +346,7 @@ func TestWrapperSuite(t *testing.T) {
t.Skip("python3 not available")
}
// the OS wrapper and (agent v0.142.0) the crash guard — both root programs in configs/ with their own suites
for _, suite := range []string{"../../configs/test_felhom_os_apply.py", "../../configs/test_felhom_crash_guard.py"} {
for _, suite := range []string{"../../configs/test_felhom_os_apply.py", "../../configs/test_felhom_crash_guard.py", "../../configs/test_felhom_config_bundle.py"} {
cmd := exec.Command(py, "-B", suite)
out, err := cmd.CombinedOutput()
if err != nil {
+5 -1
View File
@@ -51,6 +51,10 @@ const (
// A Docker engine step in the customer guest (agent v0.142.0, `11` §5.8) — ring 1, and every undo. Destructive-class
// (signed, operational key) like agent_update; the root wrapper re-verifies the same signature itself.
ClassOSDockerStep OpClass = "os_docker_step"
// The config bundle (agent v0.143.0, R-840, `11` §5.4.2): the box's ROOT-OWNED files (sudoers, wrappers, units).
// Destructive-class (signed, operational key) like agent_update; the root wrapper re-verifies the signature itself.
ClassAgentConfigUpdate OpClass = "agent_config_update"
)
// Disposition is the classifier verdict.
@@ -119,7 +123,7 @@ func Classify(class OpClass, prov Provenance) Disposition {
return Destructive
case ClassKeyRotation:
return Destructive
case ClassAgentUpdate, ClassOSDockerStep:
case ClassAgentUpdate, ClassOSDockerStep, ClassAgentConfigUpdate:
// Never benign — no agent-internal provenance can make replacing the agent binary
// unsigned-safe (a compromised process must not be able to self-bless an update).
return Destructive
+60
View File
@@ -0,0 +1,60 @@
#!/usr/bin/env python3
"""build-config-bundle.py — build the agent's CONFIG BUNDLE (R-840, `11` §5.4.2).
Usage: python3 scripts/build-config-bundle.py <agent-version> <out.json> (prints the bundle's sha256)
The bundle is every root-owned file the installer's step 5 writes for the agent (sudoers, wrappers, units), as ONE
JSON file published beside the binary (felhom-agent/<version>/felhom-config-bundle.json). A box takes it by a signed
`agent_config_update` job; a new box takes the SAME file from the installer. The list of files is NOT kept here: it is
`BUNDLE_FILES` in configs/felhom-os-apply, the root wrapper that installs it — one table, so the builder cannot put in a
path the wrapper would refuse, nor leave out one it expects.
Reproducible by construction: no timestamps, sorted keys, the table's order. The same source at the same version gives
the same sha256 every time (pinned by configs/test_felhom_config_bundle.py).
"""
import base64
import hashlib
import importlib.machinery
import importlib.util
import json
import pathlib
import re
import sys
REPO = pathlib.Path(__file__).resolve().parent.parent
CONFIGS = REPO / "configs"
def load_wrapper(configs=CONFIGS):
loader = importlib.machinery.SourceFileLoader("osapply_for_bundle", str(configs / "felhom-os-apply"))
spec = importlib.util.spec_from_loader("osapply_for_bundle", loader)
mod = importlib.util.module_from_spec(spec)
loader.exec_module(mod)
return mod
def build(version, configs=CONFIGS):
if not re.match(r"^[0-9]+\.[0-9]+\.[0-9]+(-[0-9A-Za-z.]+)?$", version):
raise SystemExit(f"build-config-bundle: version {version!r} is not semver")
w = load_wrapper(configs)
files = []
for dest, src, mode, check, policy in w.BUNDLE_FILES:
data = (configs / src).read_bytes()
files.append({"path": dest, "source": f"configs/{src}", "mode": oct(mode), "check": check, "policy": policy,
"sha256": hashlib.sha256(data).hexdigest(), "content_b64": base64.b64encode(data).decode()})
body = {"format": w.BUNDLE_FORMAT, "agent_version": version, "files": files}
return (json.dumps(body, indent=1, sort_keys=True) + "\n").encode()
def main(argv):
if len(argv) != 3:
print(__doc__, file=sys.stderr)
return 2
data = build(argv[1])
pathlib.Path(argv[2]).write_bytes(data)
print(hashlib.sha256(data).hexdigest())
return 0
if __name__ == "__main__":
sys.exit(main(sys.argv))
+21
View File
@@ -100,6 +100,15 @@ built_ver="$("$BIN" --version 2>/dev/null | awk '{print $2}')"
BUILT_SHA="$(sha256sum "$BIN" | awk '{print $1}')"
log "built ok: sha256 $BUILT_SHA"
# ── 3b. The config bundle (R-840) ───────────────────────────────────────────────────────────────
# Every root-owned file the installer's step 5 writes (sudoers, wrappers, units), as ONE file beside the binary. A box
# takes it by a signed agent_config_update; a new box from the installer. Reproducible: same source → same sha.
BUNDLE="$(mktemp -t felhom-config-bundle-XXXXXX)"
trap 'rm -f "$BIN" "$BUNDLE"' EXIT
BUNDLE_SHA="$(python3 "$REPO_ROOT/scripts/build-config-bundle.py" "$VERSION" "$BUNDLE")" || die "config bundle build failed"
[[ "$BUNDLE_SHA" =~ ^[0-9a-f]{64}$ ]] || die "config bundle build printed no sha256"
log "config bundle built: sha256 $BUNDLE_SHA"
# ── 4. Tag LOCALLY (the push comes after the publish — see step 6) ──────────────────────────────
#
# THE ORDER CHANGED, AND ONLY THE PUSH MOVED (R-188, 2026-08-03).
@@ -153,6 +162,12 @@ if ! bash "$REPO_ROOT/scripts/publish-agent.sh" "$VERSION" "$BIN"; then
die "publish failed"
fi
# ── 5b. Publish the config bundle beside the binary (same package version, same credentials) ──
BURL="$GITEA_BASE/api/packages/$GITEA_OWNER/generic/felhom-agent/$VERSION/felhom-config-bundle.json"
bcode="$(curl -sS -o /dev/null -w '%{http_code}' -u "${GITEA_USER}:${GITEA_TOKEN}" -X PUT --upload-file "$BUNDLE" "$BURL")"
[[ "$bcode" == "201" || "$bcode" == "200" ]] || die "config bundle upload failed: HTTP $bcode (the binary IS published; re-run only the bundle upload)"
log "config bundle published (HTTP $bcode)"
# ── 6. Push the tag, now that the package exists ────────────────────────────────────────────────
# This is the step that makes the release VISIBLE — to CI, and to every `raw/tag/v<version>/` fetch
# the installer makes. It runs last of the two so CI can never see a tag whose package is not there.
@@ -193,6 +208,11 @@ DL_SHA="$(sha256sum "$DL" | awk '{print $1}')"
[[ "$DL_SHA" == "$BUILT_SHA" ]] \
|| die "published sha $DL_SHA != built sha $BUILT_SHA — the artifact is not what was built"
BDL="$(mktemp -t felhom-config-bundle-dl-XXXXXX)"
trap 'rm -f "$BIN" "$DL" "$BUNDLE" "$BDL"' EXIT
curl -fsS -o "$BDL" "$BURL" || die "round-trip GET of the config bundle failed"
[[ "$(sha256sum "$BDL" | awk '{print $1}')" == "$BUNDLE_SHA" ]] || die "published config bundle sha != built sha"
# The tag must also serve the configs the installer will fetch from it.
cfg_code="$(curl -fsS -o /dev/null -w '%{http_code}' \
"$GITEA_BASE/$GITEA_OWNER/felhom-agent/raw/tag/$TAG/configs/felhom-agent.service" 2>/dev/null || true)"
@@ -206,6 +226,7 @@ cat <<EOF
version : $VERSION
tag : $TAG
sha256 : $BUILT_SHA
bundle : $BUNDLE_SHA (felhom-config-bundle.json — vouch it with the agent)
NOT VOUCHED. Vouching is what points machines at this version and stays your deliberate act:
hub operator UI → Configs → Day-0 artifacts. Until then boxes keep installing the previous one.