controller v0.71.0: fix guest-reboot recovery (boot-race + agent-path blocker)

Live diagnosis of drive-backed apps stuck Exited after a pct reboot pinned THREE
sub-causes, fixed together (hardening the existing processGuestBootChange, not a
parallel mechanism):

1. Agent-path blocker (live root cause): agentClient() returned "agent not
   configured" (cfg.LocalAPI.Endpoint empty), so processGuestBootChange AND the
   whole drive gate bailed at the first guard. bootstrap.json had a complete
   local_api block, but MaybeIngest returned immediately on "already configured"
   so a controller.yaml seeded before local_api existed never got the agent path.
   Fix: MaybeIngest now calls ensureLocalAPI on the already-configured path,
   merging local_api from bootstrap.json into the existing controller.yaml when
   missing (no hub re-pull, config preserved; idempotent + fail-safe).

2. Boot-race readiness gate: processGuestBootChange sampled BoundUnderParent once
   during fast startup, racing the ~18s rebind, recreated nothing, burned its
   boot-id one-shot. Fix: gate on the REAL live in-guest bind -- driveBindLive
   checks /mnt/felhom-drives/<drive> is a mountpoint in the controller's own /mnt
   rslave /proc/self/mountinfo; pollLiveBinds waits for it (bounded ~120s) before
   recreating via the normal pipeline. shouldRecreateOnBoot stays state-independent
   so stuck-Exited create-time-failure apps are included.

3. Single-shot fragility: processGuestBootChange ran only once at startup; a
   briefly-unreachable agent right after a guest reboot stranded recovery. Fix:
   driveGateLoop runs it every periodic tick too (idempotent, boot-id gated).

Tests (non-hollow, pre-fix companions, red-proofed): pollLiveBinds waits then
reports live / never-live stays absent / single early sample misses; ensureLocalAPI
merges local_api into a configured controller.yaml that lacks it / no-ops when
present. Live-accepted with repeated pct reboot 9201.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-06-16 16:12:49 +02:00
parent 25e5cb5850
commit e2de234325
5 changed files with 172 additions and 26 deletions
+37 -24
View File
@@ -1,33 +1,46 @@
## Changelog
### v0.71.0 — fix the guest-reboot boot-race that strands drive-backed apps (2026-06-16)
### v0.71.0 — fix guest-reboot recovery of drive-backed apps (boot-race + the agent-path blocker) (2026-06-16)
A `pct reboot` of the guest left drive-backed apps (audiobookshelf, calibre-web, immich-server,
jellyfin, komga, radarr, romm, paperless-webserver) stuck `Exited` forever. **Sub-cause (diagnosed,
not guessed):** on guest boot the in-guest dockerd auto-starts the `unless-stopped` apps **~18s before**
the agent re-binds the drive under the stable parent; the create-time volume bind fails
(`mkdir /mnt/felhom-drives/<drive>/userdata: permission denied` on the empty fail-closed placeholder)
and, being a create-time failure (`RestartCount=0`), is **never retried**. The existing recovery,
`processGuestBootChange`, *ran* but **raced the rebind**: it sampled the agent's `BoundUnderParent`
**once** during fast controller startup (bind not live yet) → recreated nothing → **persisted the new
boot-id**, burning its one-shot. The periodic drive-gate never recovered them either (its first
observation was *after* the rebind: present + not-disconnected → no transition).
jellyfin, komga, radarr, romm, paperless-webserver) stuck `Exited` forever. On guest boot the in-guest
dockerd auto-starts the `unless-stopped` apps **~18s before** the agent re-binds the drive under the
stable parent; the create-time volume bind fails (`mkdir /mnt/felhom-drives/<drive>/userdata:
permission denied` on the empty fail-closed placeholder) and, being a create-time failure
(`RestartCount=0`), is **never retried**. The intended recovery (`processGuestBootChange`) did not fire.
Live diagnosis pinned **three** sub-causes, fixed together (harden the existing mechanism — no parallel
one):
**Fix (harden the existing mechanism — no parallel one):** `processGuestBootChange` now gates on the
**REAL live in-guest bind** instead of the once-sampled agent view. New `driveBindLive` checks whether
`/mnt/felhom-drives/<drive>` is an actual mountpoint in the controller's own `/mnt` (rslave)
`/proc/self/mountinfo` — true only once the agent's bind has propagated, exactly when docker can
recreate the app. New `pollLiveBinds` waits for that (bounded ~120s, polling 2s; the rebind lands ~18s)
and only then recreates the deployed drive-backed apps via the normal pipeline (`compose down``up -d`).
`shouldRecreateOnBoot` is unchanged and state-independent, so a stuck-`Exited` create-time-failure app
is included. Single-flight (runs once before the periodic gate); apps on a drive that never goes live
in the window are left to the gate (drive-absent → stop→return→restart). The host-reboot path the
earlier sweep validated is unaffected (same code path, now strictly more robust — it waits for the
bind). The **guest-only reboot path** (which the host-reboot sweep never exercised) is now covered.
1. **The agent-path blocker (the live root cause).** `agentClient()` returned **"agent not configured"**
`cfg.LocalAPI.Endpoint` was empty — so `processGuestBootChange` (and the **entire** drive gate)
bailed at its first guard, never reaching any boot-id/bind logic. `bootstrap.json` *had* a complete
`local_api` block, but `MaybeIngest` returned immediately on "already configured" (customer.id set),
so a controller.yaml seeded before `local_api` existed never got the agent path merged. **Fix:**
`MaybeIngest` now calls new **`ensureLocalAPI`** on the already-configured path — it merges `local_api`
from bootstrap.json into the existing controller.yaml when missing (no hub re-pull, config preserved),
idempotent + fail-safe.
2. **The boot-race readiness gate.** `processGuestBootChange` sampled the agent's `BoundUnderParent`
**once** during fast startup, racing the ~18s rebind, recreated nothing, and burned its boot-id
one-shot. **Fix:** it now gates on the **REAL live in-guest bind** new `driveBindLive` checks
whether `/mnt/felhom-drives/<drive>` is an actual mountpoint in the controller's own `/mnt` (rslave)
`/proc/self/mountinfo` (true only once the agent's bind propagated, exactly when docker can recreate
the app), and new `pollLiveBinds` **waits** for it (bounded ~120s, poll 2s) before recreating via the
normal pipeline (`compose down``up -d`). `shouldRecreateOnBoot` is unchanged and state-independent,
so a stuck-`Exited` create-time-failure app is included.
3. **The single-shot fragility.** `processGuestBootChange` ran only once at startup; right after a guest
reboot the agent's local API can be briefly unreachable/stale, so the one attempt bailed and never
retried. **Fix:** `driveGateLoop` now runs it on every periodic tick too — idempotent (boot-id
gated), so it retries until the agent is reachable.
Tests: `pollLiveBinds` waits through the rebind window then reports live (recreate fires);
never-live drive stays absent (no spurious recreate); plus an explicit pre-fix companion that a single
early sample misses the not-yet-live bind. Live-accepted with repeated `pct reboot 9201`.
Apps on a drive that never goes live in the window are left to the normal gate. The host-reboot path the
earlier sweep validated is unaffected (same code path, strictly more robust); the **guest-only reboot
path** (never exercised by host-reboot sweeps) is now covered.
Tests (non-hollow, with pre-fix companions, red-proofed): `pollLiveBinds` waits through the rebind then
reports live (recreate fires) / never-live stays absent / a single early sample misses the not-yet-live
bind; `ensureLocalAPI` merges `local_api` into an already-configured controller.yaml that lacks it
(companion: pre-fix MaybeIngest left it empty) and no-ops when already present. Live-accepted with
repeated `pct reboot 9201`.
### v0.70.0 — config-apply self-restart + geo-restriction UX fixes (2026-06-16)
+8 -1
View File
@@ -655,7 +655,14 @@ not just those with HDD data. Non-HDD apps can configure destination, method, an
> rslave `/proc/self/mountinfo`?) and **waits** for it (`pollLiveBinds`, bounded ~120s) before recreating
> — including apps stuck `Exited` with a create-time mount failure (`shouldRecreateOnBoot` is
> state-independent). The **guest-only reboot path** (which the host-reboot sweep never exercised) is now
> covered; drives that never go live in the window are left to the drive-absent gate.
> covered; drives that never go live in the window are left to the drive-absent gate. `processGuestBootChange`
> also runs on every periodic `driveGateLoop` tick now (idempotent, boot-id gated) so a momentarily-unreachable
> agent right after a guest reboot no longer permanently strands recovery.
> **Agent-path prerequisite (also v0.71.0):** the whole drive gate needs `cfg.LocalAPI.Endpoint` (the
> per-guest agent local API). `bootstrap.MaybeIngest` now calls `ensureLocalAPI` on the already-configured
> path — merging `local_api` from `bootstrap.json` into an existing controller.yaml that lacks it (seeded
> before `local_api` existed) — because without it `agentClient()` returns "agent not configured" and the
> entire gate + boot recovery silently die.
>
> **⚠️ Rebuilt on the agent-delegated disk model (v0.43.0), made ROLE-AWARE in v0.44.0, UX-polished in
> v0.45.0.** After the 8C
+50 -1
View File
@@ -99,7 +99,12 @@ func Path() string {
// - On success: writes controller.yaml (0600, atomic), reloads it, and returns the reloaded cfg.
func MaybeIngest(configPath string, cfg *config.Config, logger *log.Logger, pull PullFunc) *config.Config {
if cfg != nil && cfg.Customer.ID != "" {
return cfg // already configured — do not clobber, do not pull (idempotent)
// Already configured — do NOT re-pull the hub config or clobber controller.yaml. But STILL
// ensure the per-guest local_api block is present: a controller.yaml that was seeded/pulled
// before local_api existed (or by the setup wizard) is "configured" yet has no agent path, so
// agentClient() returns "agent not configured" and the ENTIRE drive gate + guest-reboot
// recovery silently die. ensureLocalAPI merges it in from bootstrap.json if missing.
return ensureLocalAPI(configPath, cfg, logger)
}
bpath := Path()
data, err := os.ReadFile(bpath)
@@ -184,6 +189,50 @@ func pullWithRetry(pull PullFunc, hubURL, customerID, password string, logger *l
return "", lastErr
}
// ensureLocalAPI handles an ALREADY-configured controller whose controller.yaml lacks the per-guest
// local_api block (seeded/pulled before local_api existed, or set up via the wizard). Without it the
// controller cannot reach the host agent at all (agentClient → "agent not configured"), which silently
// kills the whole drive gate + guest-reboot recovery. If bootstrap.json carries a complete local_api
// block, this merges it into the existing controller.yaml in place and reloads. Idempotent + fail-safe:
// returns cfg unchanged when local_api is already present, the bootstrap is absent/incomplete, or any
// step fails (it must never brick a configured guest).
func ensureLocalAPI(configPath string, cfg *config.Config, logger *log.Logger) *config.Config {
if cfg == nil || cfg.LocalAPI.Endpoint != "" {
return cfg // already has the agent path → nothing to do
}
data, err := os.ReadFile(Path())
if err != nil {
return cfg // no bootstrap → nothing to merge (legacy/manually-configured guest)
}
var b Bootstrap
if err := json.Unmarshal(data, &b); err != nil {
return cfg
}
if b.LocalAPI.Endpoint == "" || b.LocalAPI.Fingerprint == "" || b.LocalAPI.Token == "" {
return cfg // bootstrap has no usable local_api to merge
}
current, err := os.ReadFile(configPath)
if err != nil {
return cfg
}
merged, err := mergeLocalAPI(string(current), b.LocalAPI)
if err != nil {
logger.Printf("[WARN] bootstrap: merging local_api into existing config failed: %v — agent path stays unconfigured", err)
return cfg
}
if err := writeFileAtomic(configPath, merged); err != nil {
logger.Printf("[WARN] bootstrap: could not write %s with local_api: %v", configPath, err)
return cfg
}
reloaded, err := config.LoadPermissive(configPath)
if err != nil {
logger.Printf("[WARN] bootstrap: wrote local_api but reload failed: %v", err)
return cfg
}
logger.Printf("[INFO] bootstrap: existing config was missing local_api — merged from %s (%s); agent path now configured", Path(), b.LocalAPI.Endpoint)
return reloaded
}
// mergeLocalAPI parses the pulled controller.yaml as a generic map, sets the local_api block from the
// bootstrap (overwriting any hub-emitted placeholder), and re-marshals. local_api.enabled is NOT set
// — it defaults on once endpoint is present (config.LocalAPIConfig).
@@ -137,6 +137,77 @@ func TestMaybeIngest_DoesNotClobberConfigured_NoPull(t *testing.T) {
}
}
// FIX (v0.71.0): an ALREADY-configured controller.yaml that LACKS local_api still gets the per-guest
// agent path merged from bootstrap.json — without it agentClient() returns "agent not configured" and
// the whole drive gate + guest-reboot recovery silently die. COMPANION: the pre-fix MaybeIngest
// returned immediately on "already configured", so LocalAPI.Endpoint stayed empty (this test fails
// against that). The hub is NEVER re-pulled (the existing config is preserved verbatim).
func TestMaybeIngest_ConfiguredMissingLocalAPI_Merges(t *testing.T) {
dir := t.TempDir()
_, cfgPath := writeBootstrap(t, dir, goodBootstrapV2) // bootstrap HAS local_api
const configuredNoLocalAPI = `customer:
id: cust-8200
domain: cust8200.felhom.eu
hub:
enabled: true
url: https://hub.felhom.eu
api_key: CUSTKEY
`
if err := os.WriteFile(cfgPath, []byte(configuredNoLocalAPI), 0o600); err != nil {
t.Fatal(err)
}
existing, err := config.LoadPermissive(cfgPath)
if err != nil {
t.Fatal(err)
}
if existing.LocalAPI.Endpoint != "" {
t.Fatalf("precondition: existing config should lack local_api, got %q", existing.LocalAPI.Endpoint)
}
pulled := false
pull := func(string, string, string) (string, error) { pulled = true; return hubYAML, nil }
got := MaybeIngest(cfgPath, existing, testLogger(), pull)
if pulled {
t.Fatal("must NOT re-pull the hub config for an already-configured controller")
}
if got.LocalAPI.Endpoint != "192.168.0.162:8443" || got.LocalAPI.Token != "PERGUESTTOKEN" || got.LocalAPI.Fingerprint != "ab12" {
t.Fatalf("local_api not merged into the already-configured controller (the live boot-recovery blocker): %+v", got.LocalAPI)
}
if got.Customer.ID != "cust-8200" || got.Hub.APIKey != "CUSTKEY" {
t.Fatalf("merging local_api must preserve the existing config: %+v / %+v", got.Customer, got.Hub)
}
raw, _ := os.ReadFile(cfgPath)
if !strings.Contains(string(raw), "192.168.0.162:8443") || !strings.Contains(string(raw), "CUSTKEY") {
t.Fatalf("controller.yaml must persist local_api + keep the customer key:\n%s", raw)
}
}
// IDEMPOTENT: a configured controller that ALREADY has local_api is untouched (no re-merge, no pull,
// no rewrite).
func TestMaybeIngest_ConfiguredWithLocalAPI_NoOp(t *testing.T) {
dir := t.TempDir()
_, cfgPath := writeBootstrap(t, dir, goodBootstrapV2)
existing := config.Default()
existing.Customer.ID = "cust-8200"
existing.LocalAPI.Endpoint = "already:9999"
pulled := false
pull := func(string, string, string) (string, error) { pulled = true; return hubYAML, nil }
got := MaybeIngest(cfgPath, existing, testLogger(), pull)
if pulled {
t.Fatal("must not pull when already configured")
}
if got.LocalAPI.Endpoint != "already:9999" {
t.Fatalf("existing local_api must be preserved, got %q", got.LocalAPI.Endpoint)
}
if _, err := os.Stat(cfgPath); err == nil {
t.Fatal("controller.yaml must not be rewritten when local_api already present")
}
}
// FAIL-SAFE (transient): a persistently-unreachable hub is retried, then leaves cfg in setup mode
// (no controller.yaml). Asserts the retry count (1 initial + len(pullRetryDelays)).
func TestMaybeIngest_TransientRetriesThenSetup(t *testing.T) {
+6
View File
@@ -296,6 +296,12 @@ func (s *Server) driveGateLoop() {
t := time.NewTicker(30 * time.Second)
defer t.Stop()
for range t.C {
// Re-run the boot-change converger every tick (not just once at startup): right after a GUEST
// reboot the agent's local API may briefly be unreachable / its per-guest token stale, so the
// single startup attempt can bail before reading the boot-id. It is idempotent (boot-id gated:
// a no-op once the current boot has been converged), so retrying until the agent is reachable
// is safe and is what makes guest-reboot recovery robust.
s.processGuestBootChange()
s.ReconcileDriveGates()
}
}