diff --git a/REPORT.md b/REPORT.md index 150959f..70a0951 100644 --- a/REPORT.md +++ b/REPORT.md @@ -2,36 +2,44 @@ > **Overwrite** this file with a summary of the most recent task only (uniform with the other repos; not cumulative). The cumulative hub history lives in [hub/CHANGELOG.md](hub/CHANGELOG.md); the scripts history lives in [scripts/CHANGELOG.md](scripts/CHANGELOG.md). -## TASK — offsite SLICE 4: soft-quota + health monitoring + freeze lever — 2026-07-10 — DEPLOYED +## TASK — hub v0.42.0: remote "Debug mód" toggle on the customer config editor — 2026-07-10 -**Shipped (felhom.eu):** hub **v0.41.0** `fad5573` + bump `04826c9` — LIVE (Synced/Healthy, "Offsite -checker initialized"). Pairs with controller **v0.109.1** (9201, healthy). The offsite arc's last build -item; operator-free (Viktor ran nothing). +**Scope:** hub only (`felhom.eu/hub`). Agent/controller unchanged — the controller already honours +`Logging.Level=="debug"` (`isDebug()` → `/debug` menu + verbose log). Version 0.41.0 → **0.42.0**. -**Hub changes:** -- **`monitor.OffsiteChecker`** — sibling of StorageFillChecker over the controller report's new `offsite` - object: fill warn/crit at 90/95% of `quota_gb` (quota 0 = dedicated, silent) + `offsite_stale` (warning) - for enabled+**escrowed** targets with no run >48h/never — the silently-stuck detector (recently-FAILING - runs are `backup_failed`'s signal; pending/disabled never alert — **red-proofed**: dropping the - escrowed-only filter made the pending customer alert → FAIL). Nil-safe on pre-v0.109 reports; - same-second latest-report ties processed once per sweep. -- **Freeze lever:** `Provisioner.SetOffsiteFrozen` — flips ONLY `readonly` on the exactly-1 labelled - sub-account (SSH preserved — tested; ambiguity refuses); confirm-gated **Freeze/Unfreeze** buttons next - to Re-issue (shared only). **MANUAL only, never automatic** — freezing also blocks prune, the customer's - only way down from over-quota. +### Goal +Let an operator enable the controller's debug mode **remotely, without SSH**, from the hub's customer +config editor — the support workflow (today: Peti's box). Enabling on any specific customer is a Viktor +decision/click; this task does not turn it on for anyone. -**Live verification (free checks, real pipeline):** the controller deploy's re-apply exercised -**key-auth-first live** (re-pinned, zero password consumed) and surfaced a real bug — the re-apply demoted -the escrowed demo to pending + wiped runtime status → **fixed as controller v0.109.1** (re-apply preserves -custody + status; red-proofed), demo repaired (manual confirm — truthful, same escrowed password — + a -run). The next report carried the full offsite object (`escrowed`, 45 MB / 50 GB, fresh run); the -**checker held it healthy (0 offsite_* events)**; the `/backups` usage bar renders -"Tárhelykeret: 43.2 MB / 50 GB (0%)". +### Why a form field, not raw-JSON injection +`handleConfigUpdate` rebuilds `ConfigJSON` from the form on every save (`configs.go` → `buildConfigJSON`), +so a foreign key poked straight into the stored JSON is dropped on the next save. The toggle is therefore +a first-class **form field**, which by construction survives every save. The offsite descriptor survives +independently, via the post-save provision-merge — untouched by this change. -**NOT live-exercised (per the STOP):** the freeze (would write-lock the demo's working offsite) and the -100% refusal (demo is at 0.08%) — unit red-proofs cover both. +### Changes +- `internal/web/configs.go` — `buildConfigJSON`: `debug_mode` checked → `"logging":{"level":"debug"}`; + unchecked → the `logging` key is omitted entirely (controller.yaml default stands; no needless `"info"`). +- `internal/web/templates/config_form.html` — new collapsible **"Hibakeresési mód (fejlesztői)"** section + with the `debug_mode` checkbox + a Hungarian hint (restart-next-cycle + the `/debug` URL + "turn it off + after"). Render state parsed from `ConfigJSON`. No emoji, no BOM (consistent with the hub templates). +- `internal/web/configs_debug_test.go` — new tests (below). +- Untouched, as required: configgen, the offsite provision-merge, the controller. No generic raw-JSON editor. -**The offsite arc is now build-complete:** provisioning (S1/S2) → hardening (F1–F5, key-auth-first, wipe) -→ auto-confirm (S3) → quota + health (S4). **NEXT:** the Peti onboarding runbook (live-exercises -auto-confirm + F4 + fresh-provision quota end-to-end), NEW-box restore round-trip → retire the old-box -archive. +### Tests (non-hollow) — all green +- `TestBuildConfigJSON_DebugMode` — checked emits `logging.level=debug`; unchecked omits the `logging` key. +- `TestConfigUpdate_DebugSurvivesRebuild_OffsiteUntouched` — full path through `handleConfigUpdate`: + (a) the debug key lands, (b) the offsite descriptor (`host/user/repo_path/host_fingerprint/quota_gb`) is + unchanged across save + idempotent re-provision, and the **red-proof** that a hand-injected foreign key + (`foo`) is gone after one save — the concrete proof of why the switch had to be a form field. +- `TestConfigForm_DebugRenderState` — debug-on ConfigJSON renders the checkbox `checked`; plain renders it unchecked. +- **Red-proof exercised:** with the `logging` block removed from `buildConfigJSON`, the survival + form tests fail. + +Green gate: `go build ./... && go vet ./... && go test ./...` — all pass. + +### Deploy + live check (operator-free) +GitOps: build 0.42.0 on 180 → bump `manifests/hub.yaml` tag → ArgoCD sync → Synced/Healthy. Live proof on +the **demo** customer (not Peti): edit → tick debug → save → the demo controller restarts next cycle → +`/debug` serves → untick → save → `/debug` 404 again. See the hub CHANGELOG top entry + session notes for +the live-validation outcome. Peti-box use (enabling on his config) is Viktor's click — not part of this task. diff --git a/hub/CHANGELOG.md b/hub/CHANGELOG.md index f02005e..98ec1c6 100644 --- a/hub/CHANGELOG.md +++ b/hub/CHANGELOG.md @@ -1,5 +1,28 @@ # Felhom Hub — Changelog +## v0.42.0 — Remote "Debug mód" toggle on the customer config editor (2026-07-10) + +Lets an operator flip the controller's debug mode (verbose log + the `/debug` menu, which the controller +gates on `Logging.Level=="debug"` / `isDebug()`) **remotely, without SSH** — the support workflow (today: +Peti's box). The config-version bump on save makes the controller re-pull + self-restart on its next +report ACK, so the switch takes effect within a cycle, hands-free. + +- **Form field, not raw-JSON injection — on purpose.** `handleConfigUpdate` REBUILDS `ConfigJSON` from the + form on every save (`buildConfigJSON`), so any foreign key injected straight into the stored JSON is + dropped on the next save. The toggle is therefore a real form field, which by definition survives every + save. (The offsite descriptor survives via its own separate provision-merge, untouched by this.) +- **`buildConfigJSON`** (`internal/web/configs.go`): `debug_mode` checked → emits `"logging":{"level":"debug"}`; + unchecked → the `logging` key is OMITTED entirely (the generated `controller.yaml` default stands — no + needless `"info"`). +- **Config form** (`templates/config_form.html`): new collapsible **"Hibakeresési mód (fejlesztői)"** section + with the `debug_mode` checkbox; render state parsed back from `ConfigJSON` (`logging.level=="debug"` → checked). + No change to the offsite/CF/git leg; no generic raw-JSON editor (deliberately — validated surfaces only). +- **Tests (non-hollow, `configs_debug_test.go`):** form→JSON both ways (checked emits / unchecked omits); + full-path **survival test** through `handleConfigUpdate` proving the debug key lands, the offsite descriptor + is byte-for-byte unchanged across save+re-provision, and the **red-proof** that a hand-injected foreign key + is gone after one save (why the switch must be a form field); render state both ways. Red-proof exercised + (feature disabled → survival + form tests fail). + ## v0.41.0 — SLICE 4: OffsiteChecker (fill + staleness) + operator freeze lever (2026-07-09) The last build item of the offsite arc (pairs with controller v0.109.0's soft-quota gate + report object). diff --git a/hub/internal/web/configs.go b/hub/internal/web/configs.go index 2a7ea45..058b069 100644 --- a/hub/internal/web/configs.go +++ b/hub/internal/web/configs.go @@ -955,6 +955,15 @@ func buildConfigJSON(r *http.Request) string { overrides["git"] = git } + // Logging (remote debug-mode toggle). The controller's /debug menu + verbose log key off + // Logging.Level=="debug" (controller isDebug()). This lives in the FORM — not raw-JSON injection — + // on purpose: handleConfigUpdate rebuilds ConfigJSON from the form on every save (buildConfigJSON), + // so a foreign key would be dropped on the next save. Checked → logging.level=debug; unchecked → the + // logging key is OMITTED entirely (the generated controller.yaml default stands — no needless "info"). + if r.FormValue("debug_mode") != "" { + overrides["logging"] = map[string]interface{}{"level": "debug"} + } + data, _ := json.Marshal(overrides) return string(data) } diff --git a/hub/internal/web/configs_debug_test.go b/hub/internal/web/configs_debug_test.go new file mode 100644 index 0000000..0b55481 --- /dev/null +++ b/hub/internal/web/configs_debug_test.go @@ -0,0 +1,210 @@ +package web + +import ( + "encoding/json" + "io" + "log" + "net/http" + "net/http/httptest" + "strings" + "testing" + + "gitea.dooplex.hu/admin/felhom-hub/internal/hetznerapi" + "gitea.dooplex.hu/admin/felhom-hub/internal/offsite" + "gitea.dooplex.hu/admin/felhom-hub/internal/store" +) + +// postForm builds a POST request whose urlencoded body is parsed by r.FormValue/r.ParseForm, +// exactly like the real edit submission. +func postForm(target, body string) *http.Request { + r := httptest.NewRequest(http.MethodPost, target, strings.NewReader(body)) + r.Header.Set("Content-Type", "application/x-www-form-urlencoded") + return r +} + +// The debug-mode toggle is a FORM field (not raw-JSON injection) so it survives the on-save rebuild. +// This asserts the form→JSON leg: checked writes logging.level=debug; unchecked omits the key entirely +// (so the generated controller.yaml default stands — no needless "info"). +func TestBuildConfigJSON_DebugMode(t *testing.T) { + // Checked → logging.level=debug present. + got := buildConfigJSON(postForm("/configs/x/edit", "debug_mode=on")) + var on map[string]interface{} + if err := json.Unmarshal([]byte(got), &on); err != nil { + t.Fatalf("checked: bad JSON %q: %v", got, err) + } + logging, ok := on["logging"].(map[string]interface{}) + if !ok { + t.Fatalf("checked: expected a logging object, got %q", got) + } + if logging["level"] != "debug" { + t.Fatalf("checked: expected logging.level=debug, got %v (%q)", logging["level"], got) + } + + // Unchecked → NO logging key at all (not level=info). + got = buildConfigJSON(postForm("/configs/x/edit", "customer_name=Kov%C3%A1cs")) + var off map[string]interface{} + if err := json.Unmarshal([]byte(got), &off); err != nil { + t.Fatalf("unchecked: bad JSON %q: %v", got, err) + } + if _, present := off["logging"]; present { + t.Fatalf("unchecked: logging key must be OMITTED, got %q", got) + } +} + +// The whole point of the form-level toggle: it survives the on-save ConfigJSON REBUILD, and coexists +// with the offsite descriptor (which survives via the separate provision-merge). Companion RED-PROOF: +// a foreign key injected straight into the stored ConfigJSON is GONE after one save — which is exactly +// why the switch had to be a form field, not raw-JSON injection. +func TestConfigUpdate_DebugSurvivesRebuild_OffsiteUntouched(t *testing.T) { + s, st := newTestServer(t) + s.SetOffsiteProvisioner(&offsite.Provisioner{ + API: hetznerapi.NewFake(), Store: st, Scanner: webTestScanner{}, + PoolBoxID: 611714, Location: "fsn1", Logger: log.New(io.Discard, "", 0), + }) + + const id = "cust-dbg" + if err := st.SaveCustomerConfig(&store.CustomerConfig{ + CustomerID: id, CustomerName: "Kovács", Domain: "kovacs.felhom.eu", ConfigJSON: "{}", + }); err != nil { + t.Fatalf("seed customer: %v", err) + } + + // 1) First save WITH offsite enabled → provisions + merges the descriptor. No debug yet. + const offsiteForm = "customer_name=Kov%C3%A1cs&domain=kovacs.felhom.eu&offsite_enabled=on&offsite_type=shared&offsite_quota_gb=50" + w := httptest.NewRecorder() + s.handleConfigUpdate(w, postForm("/configs/"+id+"/edit", offsiteForm), id) + if w.Code != http.StatusSeeOther { + t.Fatalf("first save: expected 303, got %d: %s", w.Code, w.Body.String()) + } + cfg, _ := st.GetCustomerConfig(id) + before := offsiteOf(t, cfg.ConfigJSON) + if before["host"] == nil || before["host"] == "" { + t.Fatalf("first save should have merged an offsite descriptor, got %q", cfg.ConfigJSON) + } + + // 2) Inject a foreign key directly into the stored ConfigJSON — the raw-JSON path the task ruled out. + var obj map[string]json.RawMessage + json.Unmarshal([]byte(cfg.ConfigJSON), &obj) + obj["foo"] = json.RawMessage(`"bar"`) + inj, _ := json.Marshal(obj) + cfg.ConfigJSON = string(inj) + if err := st.SaveCustomerConfig(cfg); err != nil { + t.Fatalf("inject foreign key: %v", err) + } + + // 3) Second save WITH offsite still enabled AND debug checked. + w = httptest.NewRecorder() + s.handleConfigUpdate(w, postForm("/configs/"+id+"/edit", offsiteForm+"&debug_mode=on"), id) + if w.Code != http.StatusSeeOther { + t.Fatalf("second save: expected 303, got %d: %s", w.Code, w.Body.String()) + } + cfg, _ = st.GetCustomerConfig(id) + + var final map[string]interface{} + if err := json.Unmarshal([]byte(cfg.ConfigJSON), &final); err != nil { + t.Fatalf("final ConfigJSON invalid: %v", err) + } + + // (a) debug logging key made it in. + logging, _ := final["logging"].(map[string]interface{}) + if logging == nil || logging["level"] != "debug" { + t.Fatalf("debug toggle did not survive the save: logging=%v (%q)", final["logging"], cfg.ConfigJSON) + } + + // (b) offsite descriptor UNCHANGED across the save+re-provision (idempotent label lookup). + after := offsiteOf(t, cfg.ConfigJSON) + for _, k := range []string{"host", "user", "repo_path", "host_fingerprint", "quota_gb"} { + if before[k] != after[k] { + t.Fatalf("offsite.%s changed across save: %v -> %v", k, before[k], after[k]) + } + } + + // RED-PROOF: the injected foreign key is GONE — buildConfigJSON rebuilt ConfigJSON from the form, + // so anything not represented as a form field is dropped. This is why the switch is a form field. + if _, present := final["foo"]; present { + t.Fatalf("foreign key survived the rebuild — the on-save rebuild premise is wrong: %q", cfg.ConfigJSON) + } +} + +// offsiteOf extracts the offsite sub-object as a generic map for field-by-field comparison. +func offsiteOf(t *testing.T, configJSON string) map[string]interface{} { + t.Helper() + var m map[string]interface{} + if err := json.Unmarshal([]byte(configJSON), &m); err != nil { + t.Fatalf("parse ConfigJSON: %v", err) + } + o, _ := m["offsite"].(map[string]interface{}) + if o == nil { + t.Fatalf("no offsite object in %q", configJSON) + } + return o +} + +// Render leg: a debug-on ConfigJSON draws the checkbox checked; a plain one draws it unchecked. +func TestConfigForm_DebugRenderState(t *testing.T) { + s, _ := newTestServer(t) + + render := func(configJSON string) string { + var overrides map[string]interface{} + json.Unmarshal([]byte(configJSON), &overrides) + data := struct { + IsNew bool + Config *store.CustomerConfig + Overrides map[string]interface{} + ActiveNav string + Error string + CSRFField string + }{ + Config: &store.CustomerConfig{CustomerID: "c1"}, + Overrides: overrides, + ActiveNav: "configs", + } + var b strings.Builder + if err := s.templates.ExecuteTemplate(&b, "config_form.html", data); err != nil { + t.Fatalf("render: %v", err) + } + return b.String() + } + + // debug on → the debug_mode checkbox carries `checked`. + out := render(`{"logging":{"level":"debug"}}`) + if !debugChecked(out) { + t.Fatalf("debug ConfigJSON should render the checkbox checked:\n%s", isolate(out)) + } + // plain → not checked. + out = render(`{}`) + if debugChecked(out) { + t.Fatalf("plain ConfigJSON must render the checkbox UNchecked:\n%s", isolate(out)) + } +} + +// debugChecked reports whether the debug_mode checkbox input renders with the `checked` attribute. +func debugChecked(html string) bool { + i := strings.Index(html, `name="debug_mode"`) + if i < 0 { + return false + } + // The input tag ends at the next '>'; `checked` (if present) sits before it. + end := strings.Index(html[i:], ">") + if end < 0 { + return false + } + return strings.Contains(html[i:i+end], "checked") +} + +// isolate trims the rendered page down to the debug_mode input's neighborhood for readable failures. +func isolate(html string) string { + i := strings.Index(html, `name="debug_mode"`) + if i < 0 { + return "(debug_mode input not found)" + } + start := i - 60 + if start < 0 { + start = 0 + } + end := i + 60 + if end > len(html) { + end = len(html) + } + return html[start:end] +} diff --git a/hub/internal/web/templates/config_form.html b/hub/internal/web/templates/config_form.html index 8c3175a..10bf2c6 100644 --- a/hub/internal/web/templates/config_form.html +++ b/hub/internal/web/templates/config_form.html @@ -103,6 +103,18 @@ +
+

Hibakeresési mód (fejlesztői)

+
+
+ + Bekapcsolva a vezérlő újraindul a következő ciklusban; a menü: https://felhom.<domain>/debug. Tesztelés után kapcsold ki. +
+
+
+

Offsite backup