src/app/routes/api/health.ts

@@ -9,8 +9,7 @@ export const Route = createFileRoute("/api/health")({
         ensureTelemetryScheduler();
         const snapshot = await getHealthSnapshot();
 
-        // Always 200: a PB/SuperTokens outage must not fail the k8s probes;
+        // Always 200: a dependency outage must not fail the k8s probes.
-        // restarting the app pod cannot fix a dependency.
         return new Response(
           JSON.stringify({
             status: snapshot.status,

src/app/routes/api/telemetry/beacon.ts

@@ -45,8 +45,6 @@ const writeEvents = async (
   const { pbAdmin } = await import("@/lib/pocketbase/client");
   await pbAdmin.authPromise;
 
-  // Sequential on purpose: SQLite is single-writer; a burst of parallel
-  // creates just queues behind the lock.
   for (const event of events) {
     try {
       if (event.kind === "error") {

src/app/start.ts

@@ -5,8 +5,6 @@ const csrfMiddleware = createCsrfMiddleware({
   filter: (ctx) => ctx.handlerType === "serverFn",
 });
 
-// Global function middleware: every server fn (current and future) is audited
-// without opting in at the definition site.
 export const startInstance = createStart(() => ({
   requestMiddleware: [csrfMiddleware],
   functionMiddleware: [serverFnLoggingMiddleware],

src/features/admin/components/dashboard/quiet-boundary.tsx

@@ -10,7 +10,6 @@ interface QuietBoundaryState {
   hasError: boolean;
 }
 
-// One failing telemetry card must not blank the whole dashboard.
 export class QuietBoundary extends Component<QuietBoundaryProps, QuietBoundaryState> {
   state: QuietBoundaryState = { hasError: false };
 

src/features/telemetry/server.ts

@@ -99,8 +99,6 @@ export const getClientErrorGroups = createServerFn()
         const sample = latest.items[0];
         if (!sample) continue;
 
-        // A group counts as resolved when its most recent occurrence is
-        // resolved; a recurrence arrives resolved=false and surfaces again.
         const resolved = !!sample.resolved;
         if (resolved && !includeResolved) continue;
 
@@ -161,8 +159,6 @@ export const getTelemetryRuntimeStatus = createServerFn()
   .middleware([superTokensAdminFunctionMiddleware])
   .handler(async () =>
     toServerResult<TelemetryRuntimeStatus>(async () => {
-      // Dev has no k8s probes hitting /api/health, so opening the dashboard
-      // is the scheduler's fallback starter.
       ensureTelemetryScheduler();
       const health = await getHealthSnapshot();
       return {

src/lib/pocketbase/services/telemetry.ts

@@ -199,8 +199,7 @@ export function createTelemetryService(pb: PocketBase) {
       };
     },
 
-    // Idempotent by the (date, metric, dim) unique index; dim must be "" (never
+    // dim must be "" (never null/undefined) or the unique index stops deduplicating.
-    // null/undefined) or the index stops deduplicating.
     async upsertRollup(input: RollupInput): Promise<RollupRecord> {
       const data = { ...input, dim: input.dim ?? "" };
       const filter = pbFilter(pb, "date = {:date} && metric = {:metric} && dim = {:dim}", {
@@ -277,8 +276,6 @@ export function createTelemetryService(pb: PocketBase) {
       };
     },
 
-    // Bounded flip of a group's rows; new occurrences arrive resolved=false by
-    // design, so a partial flip on a huge group is acceptable.
     async resolveErrorsByGroup(groupHash: string, resolved = true): Promise<number> {
       const MAX_PAGES = 25;
       let updated = 0;
@@ -300,8 +297,6 @@ export function createTelemetryService(pb: PocketBase) {
       return updated;
     },
 
-    // Paged scan over a base collection (never SQL views). Stable created sort;
-    // tight `fields` keeps payloads small for rollup passes.
     async pageCollection<T>(
       collection: string,
       opts: { filter?: string; fields?: string },

src/lib/telemetry/api-logging.server.ts

@@ -2,9 +2,6 @@ import { createMiddleware } from "@tanstack/react-start";
 import { recordActivity } from "@/utils/activities";
 import { getRequestActor } from "./request-context.server";
 
-// Request-type counterpart of serverFnLoggingMiddleware for createFileRoute
-// API handlers, which the global function middleware never sees. Not for the
-// SSE route (a long-lived stream is not a request-shaped unit) or /api/health.
 export const apiLoggingMiddleware = createMiddleware({
   type: "request",
 }).server(async ({ next, request }) => {

src/lib/telemetry/auth-events.server.ts

@@ -9,8 +9,6 @@ interface AuthEventInput {
   error?: string;
 }
 
-// Fire-and-forget: auth flows must never fail or slow down because telemetry
-// did. The phone number is used only to resolve the player, never stored.
 export const recordAuthEvent = (name: string, input: AuthEventInput) => {
   void (async () => {
     try {
@@ -52,8 +50,6 @@ export const recordAuthEvent = (name: string, input: AuthEventInput) => {
 
 type PasswordlessOverride = (originalImplementation: any) => any;
 
-// Composes with a recipe's existing override.functions so telemetry wraps the
-// final implementation (Twilio verify included).
 export const withPasswordlessTelemetry =
   (existing?: PasswordlessOverride): PasswordlessOverride =>
   (originalImplementation: any) => {

src/lib/telemetry/client/index.ts

@@ -1,10 +1,5 @@
 import type { AnyRouter } from "@tanstack/react-router";
 
-// Provider-free by design: renders nothing and mounts no React context, so it
-// cannot disturb the LinguiProvider -> AuthProvider order. Browser-only; every
-// entry point is a no-op during SSR. Runs in dev too - events go to the same
-// PB the app is pointed at, which in dev is the local instance.
-
 type QueuedEvent = Record<string, unknown> & { kind: "event" | "error" | "vital" };
 
 const FLUSH_INTERVAL_MS = 10_000;
@@ -148,8 +143,7 @@ export const initTelemetry = (router: AnyRouter) => {
     captureError("unhandledrejection", event.reason);
   });
 
-  // Vitals mostly report at hidden-visibility; their callbacks are registered
+  // Vitals report at hidden-visibility; register before the flush listeners below.
-  // before the flush listeners so those reports make the final beacon.
   void import("web-vitals")
     .then(({ onLCP, onCLS, onINP, onFCP, onTTFB }) => {
       const report = (metric: { name: string; value: number; rating: string }) => {

src/lib/telemetry/error-hash.ts

@@ -1,6 +1,3 @@
-// Deterministic client-error grouping across deploys: volatile fragments
-// (ids, URLs, asset content hashes) are normalized away before hashing.
-
 const normalizeMessage = (message: string): string =>
   message
     .toLowerCase()
@@ -13,8 +10,6 @@ const normalizeMessage = (message: string): string =>
 
 const FRAME_NOISE = /^(at\s+)?(async\s+)?(module code|<anonymous>|eval)/i;
 
-// Vite emits content-hashed asset names (index-Bq2z1a3f.js); strip the hash so
-// the same frame groups across deploys.
 const normalizeFrame = (frame: string): string =>
   frame
     .trim()

src/lib/telemetry/health.server.ts

@@ -41,8 +41,6 @@ const runChecks = async (): Promise<HealthSnapshot> => {
   return cached;
 };
 
-// Stale-while-revalidate: k8s probes get the cached snapshot instantly while a
-// refresh runs in the background; only the very first request awaits a check.
 export const getHealthSnapshot = async (): Promise<HealthSnapshot> => {
   if (cached && Date.now() - Date.parse(cached.checkedAt) < CHECK_TTL_MS) {
     return cached;

src/lib/telemetry/request-context.server.ts

@@ -1,6 +1,3 @@
-// Per-request side-channel between the session layer and the global logging
-// middleware, which runs outside session middleware and cannot see its context.
-
 const actorByRequest = new WeakMap<Request, { playerId?: string }>();
 const deniedRecorded = new WeakSet<Request>();
 

src/lib/telemetry/rollup-core.ts

@@ -20,7 +20,6 @@ export interface ClientErrorRow {
   group_hash: string;
 }
 
-// Nearest-rank percentile; values need not be pre-sorted.
 export const percentile = (values: number[], p: number): number => {
   if (values.length === 0) return 0;
   const sorted = [...values].sort((a, b) => a - b);

src/lib/telemetry/rollup.server.ts

@@ -23,8 +23,6 @@ const addDays = (dateStr: string, days: number): string => {
   return utcDay(d);
 };
 
-// PB stores autodates as "YYYY-MM-DD HH:MM:SS.sssZ" UTC; boundaries in the
-// same shape compare correctly in filters.
 const bound = (dateStr: string): string => `${dateStr} 00:00:00.000Z`;
 
 const dayFilter = (dateStr: string): string =>
@@ -91,8 +89,6 @@ export const runRollups = async (dateStr: string): Promise<number> => {
   return rollups.length;
 };
 
-// The first run after UTC midnight recomputes yesterday once more so its
-// partial rows become final. Idempotent via the (date, metric, dim) key.
 export const runScheduledRollups = async (lastRunDay?: string): Promise<string> => {
   const today = utcDay(new Date());
   if (lastRunDay && lastRunDay !== today) {

src/lib/telemetry/scheduler.server.ts

@@ -52,9 +52,7 @@ const tickAlerts = async () => {
   }
 };
 
-// Single-replica deployment makes in-process intervals a valid scheduler
+// In-process intervals are valid only while the deployment runs a single replica.
-// (same precedent as the badge migration job). Started lazily from request
-// handlers (health probes guarantee a call within seconds of boot).
 export const ensureTelemetryScheduler = () => {
   if (initialized || typeof window !== "undefined") return;
   initialized = true;

src/utils/activities.ts

@@ -41,8 +41,6 @@ const serverFnName = (request: Request, meta?: ServerFnMeta): string => {
   return pathParts[pathParts.length - 1] || "unknown";
 };
 
-// Successful reads are skipped (client page views cover usage); read failures
-// are always recorded. Naming is uniform across every feature server.ts.
 export const READ_FN_NAME = /^(get|list|search|fetch|preview|is[A-Z])/;
 
 const readSampleRate = (): number => {
@@ -56,8 +54,6 @@ const shouldRecordSuccess = (name: string): boolean => {
   return sample > 0 && Math.random() < sample;
 };
 
-// Thrown Responses/redirects are control flow (session refresh 401s, route
-// redirects), not failures.
 const isControlFlowError = (error: unknown): boolean =>
   error instanceof Response || isRedirect(error);
 
@@ -86,8 +82,7 @@ export const serverFnLoggingMiddleware = createMiddleware({
   const userAgent = request.headers.get("user-agent") || undefined;
   const safeArgs = redactArguments(data);
 
-  // Registered globally, so session middleware runs inside this one; the actor
+  // Runs outside session middleware; the actor is stashed on the Request and read after next().
-  // is stashed on the Request by getSessionContextCore and read after next().
   const resolveActor = (): string | undefined =>
     getRequestActor(request)?.playerId ??
     ((context as any)?.metadata?.player_id as string | undefined);

src/utils/supertokens.ts

@@ -26,9 +26,6 @@ export const superTokensRequestMiddleware = createMiddleware({
   return next({ context });
 });
 
-// Session context when available, anonymous otherwise — never throws or
-// redirects. For endpoints that must accept logged-out traffic (telemetry
-// beacon: login-page errors are exactly the signal we want).
 export const superTokensOptionalRequestMiddleware = createMiddleware({
   type: "request",
 }).server(async ({ next, request }) => {