Merge pull request #1235 from J3vb/feat/logging-hardening

feat: logging & error-visibility hardening (server + client)

Client/tauri-client/src-tauri/Cargo.lock

@@ -32,6 +32,23 @@ dependencies = [
  "alloc-no-stdlib",
 ]
 
+[[package]]
+name = "android_log-sys"
+version = "0.3.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "84521a3cf562bc62942e294181d9eef17eb38ceb8c68677bc49f144e4c3d4f8d"
+
+[[package]]
+name = "android_logger"
+version = "0.15.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "dbb4e440d04be07da1f1bf44fb4495ebd58669372fe0cffa6e48595ac5bd88a3"
+dependencies = [
+ "android_log-sys",
+ "env_filter",
+ "log",
+]
+
 [[package]]
 name = "android_system_properties"
 version = "0.1.5"
@@ -1256,27 +1273,14 @@ dependencies = [
 
 [[package]]
 name = "env_filter"
-version = "2.0.0"
+version = "0.1.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "900d271a03799a1ee8d1ca9b19893b48ca674a9284fefcfb85f05e74ed314217"
+checksum = "1bf3c259d255ca70051b30e2e95b5446cdb8949ac4cd22c0d7fd634d89f568e2"
 dependencies = [
  "log",
  "regex",
 ]
 
-[[package]]
-name = "env_logger"
-version = "0.11.11"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "de671bd27a75a797dc9ae289ba1e77276e75e2026408aab65185384e2d5cd3f6"
-dependencies = [
- "anstream",
- "anstyle",
- "env_filter",
- "jiff",
- "log",
-]
-
 [[package]]
 name = "equivalent"
 version = "1.0.2"
@@ -1340,6 +1344,15 @@ dependencies = [
  "simd-adler32",
 ]
 
+[[package]]
+name = "fern"
+version = "0.7.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4316185f709b23713e41e3195f90edef7fb00c3ed4adc79769cf09cc762a3b29"
+dependencies = [
+ "log",
+]
+
 [[package]]
 name = "field-offset"
 version = "0.3.6"
@@ -2350,30 +2363,6 @@ dependencies = [
  "system-deps",
 ]
 
-[[package]]
-name = "jiff"
-version = "0.2.23"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1a3546dc96b6d42c5f24902af9e2538e82e39ad350b0c766eb3fbf2d8f3d8359"
-dependencies = [
- "jiff-static",
- "log",
- "portable-atomic",
- "portable-atomic-util",
- "serde_core",
-]
-
-[[package]]
-name = "jiff-static"
-version = "0.2.23"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2a8c8b344124222efd714b73bb41f8b5120b27a7cc1c75593a6ff768d9d05aa4"
-dependencies = [
- "proc-macro2",
- "quote",
- "syn 2.0.117",
-]
-
 [[package]]
 name = "jni"
 version = "0.21.1"
@@ -2802,6 +2791,15 @@ dependencies = [
  "syn 2.0.117",
 ]
 
+[[package]]
+name = "num_threads"
+version = "0.1.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5c7398b9c8b70908f6371f47ed36737907c87c52af34c268fed0bf0ceb92ead9"
+dependencies = [
+ "libc",
+]
+
 [[package]]
 name = "objc2"
 version = "0.6.4"
@@ -3013,7 +3011,6 @@ name = "owncord-client"
 version = "1.1.0-alpha.2"
 dependencies = [
  "device_query",
- "env_logger",
  "futures-util",
  "keyring",
  "log",
@@ -3029,6 +3026,7 @@ dependencies = [
  "tauri-plugin-dialog",
  "tauri-plugin-fs",
  "tauri-plugin-http",
+ "tauri-plugin-log",
  "tauri-plugin-notification",
  "tauri-plugin-opener",
  "tauri-plugin-process",
@@ -3431,15 +3429,6 @@ version = "1.13.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c33a9471896f1c69cecef8d20cbe2f7accd12527ce60845ff44c153bb2a21b49"
 
-[[package]]
-name = "portable-atomic-util"
-version = "0.2.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "091397be61a01d4be58e7841595bd4bfedb15f1cd54977d79b8271e94ed799a3"
-dependencies = [
- "portable-atomic",
-]
-
 [[package]]
 name = "potential_utf"
 version = "0.1.4"
@@ -5073,6 +5062,27 @@ dependencies = [
  "urlpattern",
 ]
 
+[[package]]
+name = "tauri-plugin-log"
+version = "2.9.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6792296e6f389268016c77db21ebae1fc0568f2fccf88b1ec7e2ea71330afb4c"
+dependencies = [
+ "android_logger",
+ "fern",
+ "log",
+ "objc2",
+ "objc2-foundation",
+ "serde",
+ "serde_json",
+ "serde_repr",
+ "swift-rs",
+ "tauri",
+ "tauri-plugin",
+ "thiserror 2.0.18",
+ "time",
+]
+
 [[package]]
 name = "tauri-plugin-notification"
 version = "2.3.3"
@@ -5442,7 +5452,9 @@ checksum = "743bd48c283afc0388f9b8827b976905fb217ad9e647fae3a379a9283c4def2c"
 dependencies = [
  "deranged",
  "itoa",
+ "libc",
  "num-conv",
+ "num_threads",
  "powerfmt",
  "serde_core",
  "time-core",

Client/tauri-client/src-tauri/Cargo.toml

@@ -48,7 +48,10 @@ rustls = { version = "0.23", default-features = false, features = ["ring", "std"
 webpki-roots = "1"
 ring = "0.17"
 log = "0.4"
-env_logger = "0.11"
+# Writes Rust logs to a rotating file under the OS app-log dir (alongside the
+# TS client-logs) so a shipped user can retrieve them — a release build detaches
+# the console, so stdout/stderr logging is otherwise unreachable.
+tauri-plugin-log = "2"
 keyring = "3"
 rfd = { version = "0.16", default-features = false }
 

Client/tauri-client/src-tauri/src/commands.rs

@@ -47,6 +47,13 @@ pub fn get_settings(app: tauri::AppHandle) -> Result<Value, String> {
     Ok(Value::Object(map))
 }
 
+/// Log a command failure Rust-side — so it lands in the retrievable log file —
+/// and return the same message unchanged to the JS caller.
+fn log_cmd_err(cmd: &str, msg: String) -> String {
+    log::warn!("{cmd}: {msg}");
+    msg
+}
+
 #[tauri::command]
 pub fn save_settings(app: tauri::AppHandle, key: String, value: Value) -> Result<(), String> {
     if !is_settings_key_allowed(&key) {
@@ -55,12 +62,12 @@ pub fn save_settings(app: tauri::AppHandle, key: String, value: Value) -> Result
 
     let store = app
         .store(SETTINGS_STORE)
-        .map_err(|e| format!("failed to open settings store: {e}"))?;
+        .map_err(|e| log_cmd_err("save_settings", format!("failed to open settings store: {e}")))?;
 
     store.set(&key, value);
     store
         .save()
-        .map_err(|e| format!("failed to persist settings: {e}"))?;
+        .map_err(|e| log_cmd_err("save_settings", format!("failed to persist settings: {e}")))?;
     Ok(())
 }
 
@@ -102,9 +109,9 @@ pub fn store_cert_fingerprint(
         }
     }
 
-    let store = app
-        .store(CERTS_STORE)
-        .map_err(|e| format!("failed to open certs store: {e}"))?;
+    let store = app.store(CERTS_STORE).map_err(|e| {
+        log_cmd_err("store_cert_fingerprint", format!("failed to open certs store: {e}"))
+    })?;
 
     // Capture old value before mutating so we can restore it if save fails.
     let old_value = store.get(&host);
@@ -117,7 +124,10 @@ pub fn store_cert_fingerprint(
             Some(v) => { store.set(&host, v); }
             None    => { let _ = store.delete(&host); }
         }
-        return Err(format!("failed to persist cert fingerprint: {e}"));
+        return Err(log_cmd_err(
+            "store_cert_fingerprint",
+            format!("failed to persist cert fingerprint: {e}"),
+        ));
     }
     Ok(())
 }
@@ -231,7 +241,7 @@ pub fn get_identity_pin(
         .store(IDENTITY_PINS_STORE)
         .map_err(|e| format!("failed to open identity pins store: {e}"))?;
 
-    let value = store.get(&identity_pin_key(&host, &user_id)).and_then(|v| {
+    let value = store.get(identity_pin_key(&host, &user_id)).and_then(|v| {
         if let Value::String(s) = v {
             Some(s)
         } else {

Client/tauri-client/src-tauri/src/http_proxy.rs

@@ -107,12 +107,21 @@ pub async fn start_http_proxy<R: Runtime>(
         .port();
 
     let (shutdown_tx, shutdown_rx) = tokio::sync::oneshot::channel::<()>();
-    tokio::spawn(run_proxy_loop(
+    let loop_handle = tokio::spawn(run_proxy_loop(
         app.clone(),
         listener,
         remote_host.clone(),
         shutdown_rx,
     ));
+    // Watch the loop so a panic is logged instead of vanishing silently (which
+    // would leave JS with a stale cached port and no error).
+    tokio::spawn(async move {
+        match loop_handle.await {
+            Ok(()) => info!("[http_proxy] proxy loop exited"),
+            Err(e) if e.is_panic() => error!("[http_proxy] proxy loop panicked: {e:?}"),
+            Err(e) => warn!("[http_proxy] proxy loop join error: {e:?}"),
+        }
+    });
 
     info!(
         "[http_proxy] tunnel started on 127.0.0.1:{} → {}",

Client/tauri-client/src-tauri/src/lib.rs

@@ -9,6 +9,26 @@ mod tray;
 mod update_commands;
 mod ws_proxy;
 
+/// Map the RUST_LOG env var to a global level filter for the log plugin.
+/// ponytail: only the simple global form is honoured ("debug", "info", …);
+/// per-module directives like "ws_proxy=debug" fall back to Info. Add a real
+/// parser only if per-module control is actually needed.
+fn log_level_from_env() -> log::LevelFilter {
+    match std::env::var("RUST_LOG")
+        .unwrap_or_default()
+        .trim()
+        .to_ascii_lowercase()
+        .as_str()
+    {
+        "trace" => log::LevelFilter::Trace,
+        "debug" => log::LevelFilter::Debug,
+        "warn" => log::LevelFilter::Warn,
+        "error" => log::LevelFilter::Error,
+        "off" => log::LevelFilter::Off,
+        _ => log::LevelFilter::Info,
+    }
+}
+
 // Only used by the desktop-only single-instance closure below.
 #[cfg(desktop)]
 use tauri::Manager;
@@ -31,6 +51,24 @@ pub fn run() {
     }));
 
     let builder = builder
+        // Log plugin registered early so logging is available to everything
+        // after it. Writes to stdout (dev) and a rotating file in the OS
+        // app-log dir so a shipped user — whose release build has no console —
+        // can retrieve logs.
+        .plugin(
+            tauri_plugin_log::Builder::new()
+                .target(tauri_plugin_log::Target::new(
+                    tauri_plugin_log::TargetKind::Stdout,
+                ))
+                .target(tauri_plugin_log::Target::new(
+                    tauri_plugin_log::TargetKind::LogDir {
+                        file_name: Some("owncord-client".into()),
+                    },
+                ))
+                .level(log_level_from_env())
+                .max_file_size(10_000_000) // 10 MB rolling file (default 40 KB is too small)
+                .build(),
+        )
         .plugin(tauri_plugin_store::Builder::new().build())
         .plugin(tauri_plugin_notification::init())
         .plugin(tauri_plugin_http::init())
@@ -88,14 +126,7 @@ pub fn run() {
             commands::open_devtools,
         ])
         .setup(|app| {
-            // Initialize Rust logging (controlled by RUST_LOG env var, defaults to info).
-            // try_init avoids panic if another logger (e.g. a Tauri plugin) registered first.
-            let _ = env_logger::Builder::from_env(
-                env_logger::Env::default().default_filter_or("info"),
-            )
-            .format_timestamp_millis()
-            .try_init();
-
+            // Rust logging is initialized by tauri_plugin_log (registered above).
             tray::create_tray(app.handle())?;
             Ok(())
         })

Client/tauri-client/src-tauri/src/livekit_proxy.rs

@@ -134,7 +134,15 @@ pub async fn start_livekit_proxy<R: Runtime>(
 
     let (shutdown_tx, shutdown_rx) = tokio::sync::oneshot::channel::<()>();
     let host = remote_host.clone();
-    tokio::spawn(run_proxy_loop(listener, host, fingerprint, shutdown_rx));
+    let loop_handle = tokio::spawn(run_proxy_loop(listener, host, fingerprint, shutdown_rx));
+    // Watch the loop so a panic is logged instead of vanishing silently.
+    tokio::spawn(async move {
+        match loop_handle.await {
+            Ok(()) => info!("[livekit_proxy] proxy loop exited"),
+            Err(e) if e.is_panic() => error!("[livekit_proxy] proxy loop panicked: {e:?}"),
+            Err(e) => warn!("[livekit_proxy] proxy loop join error: {e:?}"),
+        }
+    });
 
     info!("[livekit_proxy] proxy started on 127.0.0.1:{} → {}", port, remote_host);
 

Client/tauri-client/src-tauri/src/ws_proxy.rs

@@ -307,13 +307,17 @@ pub fn accept_cert_fingerprint<R: Runtime>(
         return Err("fingerprint must be SHA-256 colon-hex format (e.g. aa:bb:cc:...)".into());
     }
 
-    let store = app
-        .store(CERTS_STORE)
-        .map_err(|e| format!("failed to open certs store: {e}"))?;
+    let store = app.store(CERTS_STORE).map_err(|e| {
+        log::warn!("[ws_proxy] accept_cert_fingerprint: failed to open certs store: {e}");
+        format!("failed to open certs store: {e}")
+    })?;
 
     // Capture old value before mutating so we can restore it if save fails.
     let old_value = store.get(&host);
-    store.set(&host, Value::String(fingerprint));
+    // A pin that replaces a *different* existing fingerprint is security-
+    // significant (cert rotation — or a MITM the user just accepted).
+    let changed = matches!(&old_value, Some(Value::String(s)) if *s != fingerprint);
+    store.set(&host, Value::String(fingerprint.clone()));
     if let Err(e) = store.save() {
         // Restore previous in-memory state: put back old fingerprint if one
         // existed, or delete if there was none. Without this, the new
@@ -323,7 +327,14 @@ pub fn accept_cert_fingerprint<R: Runtime>(
             Some(v) => { store.set(&host, v); }
             None    => { let _ = store.delete(&host); }
         }
+        log::warn!("[ws_proxy] accept_cert_fingerprint: failed to persist pin for {host}: {e}");
         return Err(format!("failed to persist cert fingerprint: {e}"));
     }
+    // Fingerprints are public cert hashes — safe to log; this is the TOFU audit trail.
+    if changed {
+        log::warn!("[ws_proxy] cert pin CHANGED for {host} -> {fingerprint}");
+    } else {
+        log::info!("[ws_proxy] cert pin accepted for {host} -> {fingerprint}");
+    }
     Ok(())
 }

Client/tauri-client/src/lib/api.ts

@@ -129,6 +129,9 @@ export function createApiClient(initialConfig: ApiClientConfig, onUnauthorized?:
         status: res.status,
         code: err.error,
         message: err.message,
+        // Server echoes its request ID in this header — logging it lets a
+        // client-side failure be matched to the server's log line for it.
+        reqId: res.headers.get("x-request-id") ?? undefined,
       });
       throw new ApiClientError(res.status, err.error, err.message);
     }

Client/tauri-client/src/lib/logPersistence.ts

@@ -6,7 +6,7 @@
 
 import { appLogDir, join } from "@tauri-apps/api/path";
 import { mkdir, writeTextFile, readDir, remove, exists } from "@tauri-apps/plugin-fs";
-import { type LogEntry, addLogListener, createLogger } from "./logger";
+import { type LogEntry, addLogListener, createLogger, getLogBuffer } from "./logger";
 
 const log = createLogger("logPersistence");
 const MAX_LOG_FILES = 5;
@@ -131,7 +131,18 @@ export async function initLogPersistence(): Promise<() => void> {
     currentDate = today();
     initialized = true;
 
+    // Persist entries logged before this listener attached (the bootstrap
+    // window) so a startup-time problem lands on disk, not just in the
+    // in-memory ring. Runs synchronously right before addLogListener, so there
+    // is no gap and no double-capture.
+    for (const entry of getLogBuffer()) {
+      buffer.push(JSON.stringify(entry));
+    }
+
     const removeListener = addLogListener(onLogEntry);
+    if (buffer.length > 0) {
+      scheduleFlush();
+    }
 
     return () => {
       removeListener(); // stop receiving new entries first

Client/tauri-client/src/lib/ws.ts

@@ -220,7 +220,9 @@ export function createWsClient() {
     try {
       parsed = JSON.parse(raw) as { type?: string; payload?: unknown; id?: string; seq?: number };
     } catch {
-      log.warn("Failed to parse WS message", { data: raw });
+      // Log the size only — `raw` is the decrypted frame (chat plaintext,
+      // usernames) and this line is persisted to the on-disk log.
+      log.warn("Failed to parse WS message", { bytes: raw.length });
       return;
     }
 
@@ -489,8 +491,11 @@ export function createWsClient() {
       if (msg.includes("channel full")) {
         // Outbound channel is saturated — surface the drop to listeners so an
         // optimistic row fails with retry instead of silently losing the send.
+        // Log id + size only — a slice of `json` can contain the auth
+        // envelope's bearer token, and this line is persisted to disk.
         log.warn("ws_send: outbound channel full, message dropped (backpressure)", {
-          messagePreview: json.slice(0, 120),
+          id,
+          bytes: json.length,
         });
         notifySendFailure(id, "NETWORK");
       } else {

Client/tauri-client/src/main.ts

@@ -226,7 +226,10 @@ function runHealthChecks(
           version: health.version ?? null,
           onlineUsers: health.online_users ?? null,
         });
-      } catch {
+      } catch (err) {
+        // Record why the check failed (TLS/cert-pin/network) — otherwise a
+        // "can't connect" report has no logged cause to diagnose.
+        log.warn("health check failed", { host: profile.host, error: String(err) });
         connectPage.updateHealthStatus(profile.host, {
           status: "offline",
           latencyMs: null,

Client/tauri-client/tests/unit/log-persistence.test.ts

@@ -13,6 +13,7 @@ const {
   mockRemove,
   mockReadTextFile,
   mockAddLogListener,
+  mockGetLogBuffer,
 } = vi.hoisted(() => ({
   mockAppLogDir: vi.fn().mockResolvedValue("/mock/logs"),
   mockJoin: vi.fn((...parts: string[]) => parts.join("/")),
@@ -23,6 +24,7 @@ const {
   mockRemove: vi.fn().mockResolvedValue(undefined),
   mockReadTextFile: vi.fn().mockResolvedValue(""),
   mockAddLogListener: vi.fn(),
+  mockGetLogBuffer: vi.fn(() => [] as unknown[]),
 }));
 
 vi.mock("@tauri-apps/api/path", () => ({
@@ -41,6 +43,7 @@ vi.mock("@tauri-apps/plugin-fs", () => ({
 
 vi.mock("@lib/logger", () => ({
   addLogListener: mockAddLogListener,
+  getLogBuffer: mockGetLogBuffer,
   createLogger: () => ({
     debug: vi.fn(),
     info: vi.fn(),
@@ -102,6 +105,7 @@ describe("log persistence", () => {
     mockRemove.mockReset().mockResolvedValue(undefined);
     mockReadTextFile.mockReset().mockResolvedValue("");
     mockAddLogListener.mockReset();
+    mockGetLogBuffer.mockReset().mockReturnValue([]);
   });
 
   afterEach(() => {
@@ -148,6 +152,20 @@ describe("log persistence", () => {
       expect(typeof mockAddLogListener.mock.calls[0]![0]).toBe("function");
     });
 
+    it("persists entries buffered before init (bootstrap drain)", async () => {
+      mockGetLogBuffer.mockReturnValue([makeEntry({ message: "bootstrap-line" })]);
+      const { initLogPersistence } = await freshImport();
+      captureListener();
+      await initLogPersistence();
+
+      // The drain scheduled a flush; advance past the 2000ms debounce.
+      await vi.advanceTimersByTimeAsync(2500);
+
+      expect(mockWriteTextFile).toHaveBeenCalled();
+      const written = mockWriteTextFile.mock.calls.map((c) => String(c[1])).join("");
+      expect(written).toContain("bootstrap-line");
+    });
+
     it("returns a no-op cleanup if already initialized", async () => {
       const { initLogPersistence } = await freshImport();
       captureListener();

Client/tauri-client/tests/unit/ws.test.ts

@@ -42,6 +42,7 @@ vi.spyOn(console, "error").mockImplementation(() => {});
 
 // Import after mocks are set up
 import { createWsClient, toConnectionStatus } from "../../src/lib/ws";
+import { addLogListener, type LogEntry } from "../../src/lib/logger";
 
 /** Simulate Tauri emitting an event to JS */
 function emitTauriEvent(event: string, payload: unknown): void {
@@ -291,6 +292,23 @@ describe("WebSocket Client (Tauri proxy)", () => {
     expect(messages).toHaveLength(0);
   });
 
+  it("does not log raw frame content on parse failure (no plaintext leak)", async () => {
+    client.connect({ host: "localhost:8443", token: "t" });
+    await vi.advanceTimersByTimeAsync(10);
+    emitTauriEvent("ws-state", "open");
+
+    const entries: LogEntry[] = [];
+    const remove = addLogListener((e) => entries.push(e));
+    const secret = "SUPER_SECRET_eyJhbGciOiJIUzI1NiJ9";
+    emitTauriEvent("ws-message", secret + " not-json{{{");
+    remove();
+
+    // The decrypted frame must never reach the (on-disk-persisted) log...
+    expect(JSON.stringify(entries)).not.toContain(secret);
+    // ...but the parse failure is still recorded so it stays debuggable.
+    expect(entries.some((e) => e.message.includes("Failed to parse WS message"))).toBe(true);
+  });
+
   it("disconnect prevents reconnect", async () => {
     client.connect({ host: "localhost:8443", token: "t" });
     await vi.advanceTimersByTimeAsync(10);

Server/api/channel_handler.go

@@ -1,6 +1,7 @@
 package api
 
 import (
+	"context"
 	"errors"
 	"log/slog"
 	"net/http"
@@ -134,7 +135,7 @@ func handleGetMessages(svc *service.Services) http.HandlerFunc {
 
 		msgs, hasMore, err := svc.Messages.GetMessages(r.Context(), user.ID, channelID, before, limit)
 		if err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 
@@ -200,7 +201,7 @@ func handleSearch(svc *service.Services) http.HandlerFunc {
 				})
 				return
 			}
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 		if results == nil {
@@ -232,7 +233,7 @@ func handleGetPins(svc *service.Services) http.HandlerFunc {
 
 		msgs, err := svc.Messages.GetPinnedMessages(r.Context(), user.ID, channelID)
 		if err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 
@@ -265,7 +266,7 @@ func handleSetPinned(svc *service.Services, pinned bool) http.HandlerFunc {
 		}
 
 		if err := svc.Messages.SetMessagePinned(r.Context(), user.ID, channelID, messageID, pinned); err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 		w.WriteHeader(http.StatusNoContent)
@@ -273,7 +274,7 @@ func handleSetPinned(svc *service.Services, pinned bool) http.HandlerFunc {
 }
 
 // writeServiceError maps a service-layer error to an HTTP response.
-func writeServiceError(w http.ResponseWriter, err error) {
+func writeServiceError(ctx context.Context, w http.ResponseWriter, err error) {
 	switch {
 	case errors.Is(err, service.ErrRateLimited):
 		writeJSON(w, http.StatusTooManyRequests, errorResponse{Error: "RATE_LIMITED", Message: err.Error()})
@@ -286,10 +287,12 @@ func writeServiceError(w http.ResponseWriter, err error) {
 	case errors.Is(err, service.ErrConflict):
 		writeJSON(w, http.StatusConflict, errorResponse{Error: "CONFLICT", Message: err.Error()})
 	case errors.Is(err, service.ErrInternal):
-		slog.Error("service error", "err", err)
+		// ErrorContext so the enriching handler attaches req_id/trace_id,
+		// linking this 500 to its request log line and trace.
+		slog.ErrorContext(ctx, "service error", "error", err)
 		writeJSON(w, http.StatusInternalServerError, errorResponse{Error: "INTERNAL_ERROR", Message: "an internal error occurred"})
 	default:
-		slog.Error("service error", "err", err)
+		slog.ErrorContext(ctx, "service error", "error", err)
 		writeJSON(w, http.StatusInternalServerError, errorResponse{Error: "INTERNAL_ERROR", Message: "internal error"})
 	}
 }

Server/api/dm_handler.go

@@ -75,7 +75,7 @@ func handleCreateDM(svc *service.Services) http.HandlerFunc {
 
 		result, err := svc.DMs.CreateDM(r.Context(), user.ID, req.RecipientID)
 		if err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 
@@ -115,7 +115,7 @@ func handleListDMs(svc *service.Services) http.HandlerFunc {
 
 		channels, err := svc.DMs.ListDMs(r.Context(), user.ID)
 		if err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 		writeJSON(w, http.StatusOK, listDMsResponse{DMChannels: channels})
@@ -139,7 +139,7 @@ func handleCloseDM(svc *service.Services, broadcaster DMBroadcaster) http.Handle
 		}
 
 		if err := svc.DMs.CloseDM(r.Context(), user.ID, channelID); err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 
@@ -170,7 +170,7 @@ func handleBlockUser(svc *service.Services) http.HandlerFunc {
 		}
 
 		if err := svc.Blocks.BlockUser(r.Context(), user.ID, targetID); err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 		writeJSON(w, http.StatusOK, map[string]string{"message": "user blocked"})
@@ -192,7 +192,7 @@ func handleUnblockUser(svc *service.Services) http.HandlerFunc {
 		}
 
 		if err := svc.Blocks.UnblockUser(r.Context(), user.ID, targetID); err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 		writeJSON(w, http.StatusOK, map[string]string{"message": "user unblocked"})
@@ -210,7 +210,7 @@ func handleListBlocks(svc *service.Services) http.HandlerFunc {
 
 		ids, err := svc.Blocks.ListBlocked(r.Context(), user.ID)
 		if err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 		writeJSON(w, http.StatusOK, map[string]any{"blocked_user_ids": ids})

Server/api/invite_handler.go

@@ -75,7 +75,7 @@ func handleCreateInvite(svc *service.Services) http.HandlerFunc {
 
 		inv, err := svc.Invites.CreateInvite(r.Context(), user.ID, req.MaxUses, req.ExpiresInHours)
 		if err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 		writeJSON(w, http.StatusCreated, toInviteResponse(inv))
@@ -87,7 +87,7 @@ func handleListInvites(svc *service.Services) http.HandlerFunc {
 	return func(w http.ResponseWriter, r *http.Request) {
 		invites, err := svc.Invites.ListInvites(r.Context())
 		if err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 
@@ -104,7 +104,7 @@ func handleRevokeInvite(svc *service.Services) http.HandlerFunc {
 	return func(w http.ResponseWriter, r *http.Request) {
 		code := chi.URLParam(r, "code")
 		if err := svc.Invites.RevokeInvite(r.Context(), code); err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 		w.WriteHeader(http.StatusNoContent)

Server/api/middleware.go

@@ -44,6 +44,11 @@ func AuthMiddleware(database *db.DB) func(http.Handler) http.Handler {
 			hash := auth.HashToken(token)
 			sess, err := database.GetSessionByTokenHash(r.Context(), hash)
 			if err != nil || sess == nil {
+				if err != nil {
+					// A DB error here is an outage, not a bad token — log it so
+					// it's distinguishable from ordinary invalid-token 401s.
+					slog.ErrorContext(r.Context(), "auth: session lookup failed", "error", err)
+				}
 				writeJSON(w, http.StatusUnauthorized, errorResponse{
 					Error:   "UNAUTHORIZED",
 					Message: "invalid or expired session",
@@ -58,7 +63,9 @@ func AuthMiddleware(database *db.DB) func(http.Handler) http.Handler {
 				// written, so detach cancellation: the deletion must complete.
 				cleanupCtx := context.WithoutCancel(r.Context())
 				go func(h string) {
-					_ = database.DeleteSession(cleanupCtx, h)
+					if err := database.DeleteSession(cleanupCtx, h); err != nil {
+						slog.WarnContext(cleanupCtx, "expired session cleanup failed", "error", err)
+					}
 				}(hash)
 				writeJSON(w, http.StatusUnauthorized, errorResponse{
 					Error:   "UNAUTHORIZED",
@@ -70,6 +77,9 @@ func AuthMiddleware(database *db.DB) func(http.Handler) http.Handler {
 			// Load user.
 			user, err := database.GetUserByID(r.Context(), sess.UserID)
 			if err != nil || user == nil {
+				if err != nil {
+					slog.ErrorContext(r.Context(), "auth: user lookup failed", "error", err, "user_id", sess.UserID)
+				}
 				writeJSON(w, http.StatusUnauthorized, errorResponse{
 					Error:   "UNAUTHORIZED",
 					Message: "user not found",
@@ -92,6 +102,9 @@ func AuthMiddleware(database *db.DB) func(http.Handler) http.Handler {
 			// and every downstream permission check has to re-guard it.
 			role, err := database.GetRoleByID(r.Context(), user.RoleID)
 			if err != nil || role == nil {
+				if err != nil {
+					slog.ErrorContext(r.Context(), "auth: role lookup failed", "error", err, "user_id", user.ID, "role_id", user.RoleID)
+				}
 				writeJSON(w, http.StatusUnauthorized, errorResponse{
 					Error:   "UNAUTHORIZED",
 					Message: "role not found",

Server/api/profile_handler.go

@@ -171,14 +171,14 @@ func handleUpdateProfile(svc *service.Services, broadcaster ProfileBroadcaster)
 
 		updated, err := svc.Users.UpdateProfile(r.Context(), user.ID, req.Username, req.Avatar)
 		if err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 
 		if req.IdentityPublicKey != nil {
 			updated, err = svc.Users.UpdateIdentityKey(r.Context(), user.ID, *req.IdentityPublicKey)
 			if err != nil {
-				writeServiceError(w, err)
+				writeServiceError(r.Context(), w, err)
 				return
 			}
 		}
@@ -275,7 +275,7 @@ func handleChangePassword(svc *service.Services, limiter *auth.RateLimiter) http
 		res, err := svc.Users.ChangePassword(r.Context(), user.ID, hash, keepSessionID)
 		if err != nil {
 			// Only reachable when the password itself failed to commit.
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 		if res.RevokeFailed {
@@ -314,7 +314,7 @@ func handleListSessions(svc *service.Services) http.HandlerFunc {
 
 		sessions, err := svc.Users.ListSessions(r.Context(), user.ID)
 		if err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 
@@ -353,7 +353,7 @@ func handleRevokeSession(svc *service.Services) http.HandlerFunc {
 		}
 
 		if err := svc.Users.RevokeSession(r.Context(), user.ID, sessionID); err != nil {
-			writeServiceError(w, err)
+			writeServiceError(r.Context(), w, err)
 			return
 		}
 

Server/api/router.go

@@ -18,6 +18,7 @@ import (
 	"github.com/owncord/server/permissions"
 	"github.com/owncord/server/plugin"
 	"github.com/owncord/server/service"
+	"github.com/owncord/server/stackutil"
 	"github.com/owncord/server/storage"
 	"github.com/owncord/server/telemetry"
 	"github.com/owncord/server/updater"
@@ -39,7 +40,7 @@ func NewRouter(cfg *config.Config, database *db.DB, ver string, logBuf *admin.Ri
 	// NOTE: middleware.RealIP is intentionally omitted — trusting X-Real-IP from
 	// any source allows IP spoofing for rate-limit bypass. IP header trust is now
 	// handled explicitly in clientIPWithProxies using the trusted_proxies config.
-	r.Use(middleware.Recoverer)
+	r.Use(recoverer)     // slog-routing panic recovery (replaces chi's stderr-only Recoverer)
 	r.Use(requestLogger) // structured request/response logging
 	// Phase B Step 8 — OpenTelemetry HTTP tracing. No-op when telemetry is
 	// disabled or the otel build tag is not set, so this is safe to mount
@@ -363,6 +364,44 @@ func setRequestIDHeader(next http.Handler) http.Handler {
 	})
 }
 
+// recoverer recovers from panics in HTTP handlers and logs them through slog —
+// so they reach the admin log stream and are structured — unlike chi's default
+// middleware.Recoverer, which writes an unstructured stack to stderr only. The
+// stack is captured via stackutil so it never embeds argument values (which on
+// auth/upload paths can carry tokens or passwords).
+func recoverer(next http.Handler) http.Handler {
+	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		// Capture correlation IDs before dispatch so the recovery closure makes
+		// no context calls (which trip contextcheck inside a defer), while the
+		// panic log still carries req_id/trace_id.
+		reqID := middleware.GetReqID(r.Context())
+		traceID := telemetry.TraceIDFromContext(r.Context())
+		defer func() {
+			if rec := recover(); rec != nil {
+				// Preserve chi's behaviour of not swallowing the abort sentinel.
+				if rec == http.ErrAbortHandler {
+					panic(rec)
+				}
+				attrs := []any{
+					"method", r.Method,
+					"path", r.URL.Path,
+					"panic", rec,
+					"stack", stackutil.Capture(),
+				}
+				if reqID != "" {
+					attrs = append(attrs, "req_id", reqID)
+				}
+				if traceID != "" {
+					attrs = append(attrs, "trace_id", traceID)
+				}
+				slog.Error("http handler panic recovered", attrs...)
+				w.WriteHeader(http.StatusInternalServerError)
+			}
+		}()
+		next.ServeHTTP(w, r)
+	})
+}
+
 // requestLogger logs every HTTP request with method, path, status, and duration.
 // Health checks are logged at Debug level to avoid noise.
 func requestLogger(next http.Handler) http.Handler {

Server/config/config.go

@@ -30,6 +30,36 @@ type Config struct {
 	Telemetry        TelemetryConfig        `koanf:"telemetry"`
 	Plugins          PluginsConfig          `koanf:"plugins"`
 	GIF              GIFConfig              `koanf:"gif"`
+	Logging          LoggingConfig          `koanf:"logging"`
+}
+
+// LoggingConfig controls server log verbosity. The in-memory ring buffer that
+// backs the admin panel's live log view always captures DEBUG regardless of
+// this setting — Level only gates what is written to stdout.
+type LoggingConfig struct {
+	// Level is the minimum level written to stdout: "debug" | "info" | "warn" |
+	// "error". Override at runtime without editing config.yaml via the
+	// OWNCORD_LOGGING_LEVEL environment variable.
+	Level string `koanf:"level"`
+}
+
+// ParseLevel maps a config log-level string to a slog.Level. It is
+// case-insensitive and treats "" as info. The bool is false for an
+// unrecognised value (in which case slog.LevelInfo is returned and the caller
+// should warn) so a typo doesn't silently disable logging.
+func ParseLevel(s string) (slog.Level, bool) {
+	switch strings.ToLower(strings.TrimSpace(s)) {
+	case "debug":
+		return slog.LevelDebug, true
+	case "", "info":
+		return slog.LevelInfo, true
+	case "warn", "warning":
+		return slog.LevelWarn, true
+	case "error":
+		return slog.LevelError, true
+	default:
+		return slog.LevelInfo, false
+	}
 }
 
 // GIFConfig holds the credentials for the server-side GIF (Klipy) proxy.
@@ -221,6 +251,9 @@ func defaults() Config {
 			CPUBudgetMs:   100,
 			HTTPAllowlist: []string{},
 		},
+		Logging: LoggingConfig{
+			Level: "info",
+		},
 	}
 }
 
@@ -305,6 +338,12 @@ voice:
 # Get a key at https://partner.klipy.com
 # gif:
 #   api_key: ""
+
+# Logging. "level" gates what is written to stdout; the admin panel's live log
+# view always captures debug regardless. Override without editing this file via
+# the OWNCORD_LOGGING_LEVEL environment variable.
+# logging:
+#   level: "info"             # debug | info | warn | error
 `
 
 // Load reads configuration from the given YAML file path, merging with

Server/config/logging_test.go

@@ -0,0 +1,48 @@
+package config
+
+import (
+	"log/slog"
+	"path/filepath"
+	"testing"
+)
+
+func TestParseLevel(t *testing.T) {
+	cases := []struct {
+		in   string
+		want slog.Level
+		ok   bool
+	}{
+		{"debug", slog.LevelDebug, true},
+		{"info", slog.LevelInfo, true},
+		{"", slog.LevelInfo, true},
+		{"WARN", slog.LevelWarn, true},
+		{"warning", slog.LevelWarn, true},
+		{"error", slog.LevelError, true},
+		{"  Debug  ", slog.LevelDebug, true},
+		{"bogus", slog.LevelInfo, false},
+	}
+	for _, c := range cases {
+		got, ok := ParseLevel(c.in)
+		if got != c.want || ok != c.ok {
+			t.Errorf("ParseLevel(%q) = %v,%v want %v,%v", c.in, got, ok, c.want, c.ok)
+		}
+	}
+}
+
+// TestLoggingLevelFromEnv verifies the end-to-end wiring: OWNCORD_LOGGING_LEVEL
+// overrides config.yaml via koanf's existing env layer.
+func TestLoggingLevelFromEnv(t *testing.T) {
+	cfgPath := filepath.Join(t.TempDir(), "config.yaml")
+	t.Setenv("OWNCORD_LOGGING_LEVEL", "debug")
+
+	cfg, err := Load(cfgPath)
+	if err != nil {
+		t.Fatalf("Load: %v", err)
+	}
+	if cfg.Logging.Level != "debug" {
+		t.Errorf("expected level %q from env, got %q", "debug", cfg.Logging.Level)
+	}
+	if lvl, ok := ParseLevel(cfg.Logging.Level); !ok || lvl != slog.LevelDebug {
+		t.Errorf("ParseLevel(%q) = %v,%v", cfg.Logging.Level, lvl, ok)
+	}
+}

Server/config/logvalue.go

@@ -0,0 +1,69 @@
+package config
+
+import "log/slog"
+
+// This file makes secret-bearing config types safe to log by construction.
+// Without it, "never log secrets" holds only by call-site discipline — a
+// single slog.Info("cfg", "voice", cfg.Voice) would dump the LiveKit key. By
+// implementing slog.LogValuer, secrets are redacted no matter how the value
+// reaches a log record. Non-secret fields stay visible so the logs remain
+// useful for debugging.
+
+// redactSecret masks a secret for logging. An empty value stays empty (so an
+// unset credential is still visible as "unset"); anything else becomes a fixed
+// marker that reveals neither the value nor its length.
+func redactSecret(s string) string {
+	if s == "" {
+		return ""
+	}
+	return "[REDACTED]"
+}
+
+// LogValue redacts the LiveKit API key and secret.
+func (v VoiceConfig) LogValue() slog.Value {
+	return slog.GroupValue(
+		slog.String("livekit_api_key", redactSecret(v.LiveKitAPIKey)),
+		slog.String("livekit_api_secret", redactSecret(v.LiveKitAPISecret)),
+		slog.String("livekit_url", v.LiveKitURL),
+		slog.String("livekit_binary", v.LiveKitBinaryPath),
+		slog.String("node_ip", v.NodeIP),
+		slog.Bool("advertise_internal_ip", v.AdvertiseInternalIP),
+		slog.String("quality", v.Quality),
+	)
+}
+
+// LogValue redacts the GitHub token.
+func (g GitHubConfig) LogValue() slog.Value {
+	return slog.GroupValue(
+		slog.String("token", redactSecret(g.Token)),
+		slog.String("owner", g.Owner),
+		slog.String("repo", g.Repo),
+	)
+}
+
+// LogValue redacts the Klipy (GIF proxy) API key.
+func (g GIFConfig) LogValue() slog.Value {
+	return slog.GroupValue(
+		slog.String("api_key", redactSecret(g.APIKey)),
+	)
+}
+
+// LogValue delegates each section through slog.Any so secret-bearing sections
+// are redacted via their own LogValue even when the whole Config is logged.
+// A section added to Config but not listed here is omitted from the log (safe:
+// it is hidden, never leaked).
+func (c Config) LogValue() slog.Value {
+	return slog.GroupValue(
+		slog.Any("server", c.Server),
+		slog.Any("database", c.Database),
+		slog.Any("tls", c.TLS),
+		slog.Any("upload", c.Upload),
+		slog.Any("voice", c.Voice),
+		slog.Any("github", c.GitHub),
+		slog.Any("event_persistence", c.EventPersistence),
+		slog.Any("telemetry", c.Telemetry),
+		slog.Any("plugins", c.Plugins),
+		slog.Any("gif", c.GIF),
+		slog.Any("logging", c.Logging),
+	)
+}

Server/config/logvalue_test.go

@@ -0,0 +1,45 @@
+package config
+
+import (
+	"bytes"
+	"log/slog"
+	"strings"
+	"testing"
+)
+
+// TestSecretConfigsRedactedInLogs is the security guard for config redaction:
+// no matter whether the whole Config or a single section is logged, the
+// LiveKit key/secret, GitHub token, and Klipy key must never reach the output.
+func TestSecretConfigsRedactedInLogs(t *testing.T) {
+	const (
+		lkKey    = "LIVEKIT_KEY_should_not_appear"
+		lkSecret = "LIVEKIT_SECRET_should_not_appear"
+		ghToken  = "ghp_token_should_not_appear"
+		gifKey   = "klipy_key_should_not_appear"
+	)
+	cfg := Config{
+		Voice:  VoiceConfig{LiveKitAPIKey: lkKey, LiveKitAPISecret: lkSecret, Quality: "high"},
+		GitHub: GitHubConfig{Token: ghToken, Owner: "acme"},
+		GIF:    GIFConfig{APIKey: gifKey},
+	}
+
+	var buf bytes.Buffer
+	log := slog.New(slog.NewTextHandler(&buf, nil))
+	log.Info("whole", "config", cfg) // whole-config path (delegating LogValue)
+	log.Info("voice", "voice", cfg.Voice)
+	log.Info("github", "github", cfg.GitHub)
+	log.Info("gif", "gif", cfg.GIF)
+
+	out := buf.String()
+	for _, secret := range []string{lkKey, lkSecret, ghToken, gifKey} {
+		if strings.Contains(out, secret) {
+			t.Errorf("secret leaked into log output: %q\n%s", secret, out)
+		}
+	}
+	if !strings.Contains(out, "high") || !strings.Contains(out, "acme") {
+		t.Errorf("expected non-secret fields in output:\n%s", out)
+	}
+	if !strings.Contains(out, "[REDACTED]") {
+		t.Errorf("expected redaction marker in output:\n%s", out)
+	}
+}

Server/db/logvalue.go

@@ -0,0 +1,34 @@
+package db
+
+import "log/slog"
+
+// This file makes the secret-bearing domain types safe to log by construction.
+// User.PasswordHash / User.TOTPSecret and Session.TokenHash are json:"-", but
+// that only guards JSON responses — slog renders struct fields regardless. By
+// implementing slog.LogValuer, logging a *db.User or *db.Session (e.g.
+// slog.Info("x", "user", user)) never emits the credential. Secret fields are
+// omitted entirely; the useful identifying fields stay visible.
+
+// LogValue omits PasswordHash and TOTPSecret. It exposes whether TOTP is
+// enabled (not the secret) since that is often what a log line needs.
+func (u User) LogValue() slog.Value {
+	return slog.GroupValue(
+		slog.Int64("id", u.ID),
+		slog.String("username", u.Username),
+		slog.Int64("role_id", u.RoleID),
+		slog.String("status", u.Status),
+		slog.Bool("banned", u.Banned),
+		slog.Bool("totp_enabled", u.TOTPSecret != nil),
+	)
+}
+
+// LogValue omits TokenHash (the session-identifying secret).
+func (s Session) LogValue() slog.Value {
+	return slog.GroupValue(
+		slog.Int64("id", s.ID),
+		slog.Int64("user_id", s.UserID),
+		slog.String("device", s.Device),
+		slog.String("ip", s.IP),
+		slog.String("expires_at", s.ExpiresAt),
+	)
+}

Server/db/logvalue_test.go

@@ -0,0 +1,41 @@
+package db
+
+import (
+	"bytes"
+	"log/slog"
+	"strings"
+	"testing"
+)
+
+// TestUserSessionRedactedInLogs is the security guard for domain-type
+// redaction: logging a db.User / db.Session (by value or by pointer) must never
+// emit the password hash, TOTP secret, or session token hash.
+func TestUserSessionRedactedInLogs(t *testing.T) {
+	const (
+		pwHash = "PWHASH_should_not_appear"
+		totp   = "TOTP_SECRET_should_not_appear"
+		token  = "TOKENHASH_should_not_appear"
+	)
+	totpPtr := totp
+	user := User{ID: 7, Username: "alice", PasswordHash: pwHash, TOTPSecret: &totpPtr, RoleID: 2}
+	sess := Session{ID: 3, UserID: 7, TokenHash: token, Device: "cli"}
+
+	var buf bytes.Buffer
+	log := slog.New(slog.NewTextHandler(&buf, nil))
+	// Value and pointer paths — both must resolve LogValue.
+	log.Info("u", "user", user, "user_ptr", &user)
+	log.Info("s", "session", sess, "session_ptr", &sess)
+
+	out := buf.String()
+	for _, secret := range []string{pwHash, totp, token} {
+		if strings.Contains(out, secret) {
+			t.Errorf("secret leaked into log output: %q\n%s", secret, out)
+		}
+	}
+	if !strings.Contains(out, "alice") {
+		t.Errorf("expected username in output:\n%s", out)
+	}
+	if !strings.Contains(out, "totp_enabled=true") {
+		t.Errorf("expected totp_enabled flag in output:\n%s", out)
+	}
+}

Server/logctx/logctx.go

@@ -0,0 +1,54 @@
+// Package logctx provides a slog.Handler that enriches every log record with
+// correlation IDs pulled from the context: the chi request ID (as req_id) and,
+// under -tags otel, the OpenTelemetry trace ID (as trace_id).
+//
+// Wrapping the base handler with New means any log call using the ...Context
+// variants (slog.InfoContext, slog.ErrorContext, …) automatically carries
+// these IDs, so a log line can be tied back to its HTTP request and its
+// distributed trace without threading the IDs through by hand at every site.
+package logctx
+
+import (
+	"context"
+	"log/slog"
+
+	"github.com/go-chi/chi/v5/middleware"
+	"github.com/owncord/server/telemetry"
+)
+
+type handler struct {
+	inner slog.Handler
+}
+
+// New wraps inner so that records handled with a context carrying a chi request
+// ID (and, in the otel build, an active span) are enriched with req_id and
+// trace_id attributes.
+func New(inner slog.Handler) slog.Handler {
+	return handler{inner: inner}
+}
+
+func (h handler) Enabled(ctx context.Context, l slog.Level) bool {
+	return h.inner.Enabled(ctx, l)
+}
+
+func (h handler) Handle(ctx context.Context, r slog.Record) error {
+	if reqID := middleware.GetReqID(ctx); reqID != "" {
+		r.AddAttrs(slog.String("req_id", reqID))
+	}
+	if tid := telemetry.TraceIDFromContext(ctx); tid != "" {
+		r.AddAttrs(slog.String("trace_id", tid))
+	}
+	return h.inner.Handle(ctx, r)
+}
+
+func (h handler) WithAttrs(attrs []slog.Attr) slog.Handler {
+	return handler{inner: h.inner.WithAttrs(attrs)}
+}
+
+// WithGroup re-wraps so enrichment survives logger.WithGroup.
+// ponytail: req_id/trace_id are added at the record's top level; the codebase
+// opens no logger-level groups, so there is no group-nesting concern to handle
+// here. Revisit if slog group usage is introduced.
+func (h handler) WithGroup(name string) slog.Handler {
+	return handler{inner: h.inner.WithGroup(name)}
+}

Server/logctx/logctx_test.go

@@ -0,0 +1,38 @@
+package logctx
+
+import (
+	"bytes"
+	"context"
+	"log/slog"
+	"strings"
+	"testing"
+
+	"github.com/go-chi/chi/v5/middleware"
+)
+
+// TestHandlerAddsReqID verifies the enriching handler stamps req_id when the
+// context carries a chi request ID, and omits it otherwise. (trace_id is only
+// populated under -tags otel and is covered by manual verification.)
+func TestHandlerAddsReqID(t *testing.T) {
+	var buf bytes.Buffer
+	log := slog.New(New(slog.NewTextHandler(&buf, nil)))
+
+	ctx := context.WithValue(context.Background(), middleware.RequestIDKey, "req-abc-123")
+	log.InfoContext(ctx, "hello")
+	if !strings.Contains(buf.String(), "req_id=req-abc-123") {
+		t.Errorf("expected req_id in output, got: %s", buf.String())
+	}
+
+	buf.Reset()
+	log.InfoContext(context.Background(), "plain")
+	if strings.Contains(buf.String(), "req_id") {
+		t.Errorf("did not expect req_id without a request ID: %s", buf.String())
+	}
+
+	// Enrichment must survive logger.With (WithAttrs re-wrap).
+	buf.Reset()
+	log.With("k", "v").InfoContext(ctx, "withattrs")
+	if !strings.Contains(buf.String(), "req_id=req-abc-123") {
+		t.Errorf("expected req_id to survive With(): %s", buf.String())
+	}
+}

Server/main.go

@@ -23,6 +23,7 @@ import (
 	"github.com/owncord/server/auth"
 	"github.com/owncord/server/config"
 	"github.com/owncord/server/db"
+	"github.com/owncord/server/logctx"
 	"github.com/owncord/server/plugin"
 	"github.com/owncord/server/storage"
 	"github.com/owncord/server/telemetry"
@@ -36,12 +37,18 @@ func main() {
 	// Create ring buffer for admin log viewer, then build a multi-handler
 	// that tees log records to both stdout (INFO+) and the ring buffer (DEBUG+).
 	logBuf := admin.NewRingBuffer(2000)
-	stdoutHandler := slog.NewTextHandler(os.Stdout, &slog.HandlerOptions{Level: slog.LevelInfo})
+	// levelVar controls the stdout handler's threshold. It starts at INFO (the
+	// zero value) so early-startup logs are captured, then run() raises/lowers
+	// it once config.yaml / OWNCORD_LOGGING_LEVEL is loaded.
+	levelVar := new(slog.LevelVar)
+	stdoutHandler := slog.NewTextHandler(os.Stdout, &slog.HandlerOptions{Level: levelVar})
 	multiHandler := admin.NewMultiHandler(stdoutHandler, logBuf, slog.LevelDebug)
-	log := slog.New(multiHandler)
+	// logctx enriches records logged with a request/trace context (the
+	// ...Context slog variants) with req_id and, under -tags otel, trace_id.
+	log := slog.New(logctx.New(multiHandler))
 	slog.SetDefault(log)
 
-	if err := run(log, logBuf); err != nil {
+	if err := run(log, logBuf, levelVar); err != nil {
 		_, _ = fmt.Fprintf(os.Stderr, "\n  [ERROR] %v\n\n", err)
 		log.Error("server exited with error", "error", err)
 		os.Exit(1)
@@ -49,7 +56,7 @@ func main() {
 }
 
 // run is the real entrypoint — separated for testability.
-func run(log *slog.Logger, logBuf *admin.RingBuffer) error {
+func run(log *slog.Logger, logBuf *admin.RingBuffer, levelVar *slog.LevelVar) error {
 	// bgCtx is a cancellable context shared by all background goroutines
 	// (event persister, event pruner, plugin loader, maintenance loop).
 	// It is cancelled early in the shutdown sequence so in-flight DB
@@ -78,6 +85,14 @@ func run(log *slog.Logger, logBuf *admin.RingBuffer) error {
 		return fmt.Errorf("loading config: %w", err)
 	}
 
+	// Apply the configured stdout log level. The ring buffer keeps capturing
+	// DEBUG regardless, so the admin panel's live log view is unaffected.
+	if lvl, ok := config.ParseLevel(cfg.Logging.Level); ok {
+		levelVar.Set(lvl)
+	} else {
+		log.Warn("unknown logging.level, keeping info", "value", cfg.Logging.Level)
+	}
+
 	// ── 2. Ensure data directory exists ────────────────────────────────────
 	if mkdirErr := os.MkdirAll(cfg.Server.DataDir, 0o750); mkdirErr != nil {
 		return fmt.Errorf("creating data dir %s: %w", cfg.Server.DataDir, mkdirErr)
@@ -366,7 +381,7 @@ func isAddrInUse(err error) bool {
 }
 
 // printBanner writes the startup banner to stderr (so it doesn't mix with
-// JSON-structured log output on stdout).
+// the structured log output on stdout).
 func printBanner(cfg *config.Config, ver string, tls bool) {
 	scheme := "http"
 	if tls {

Server/service/block.go

@@ -46,7 +46,7 @@ func (s *BlockService) BlockUser(ctx context.Context, blockerID, targetID int64)
 	}
 
 	if err := s.st.BlockUser(ctx, blockerID, targetID); err != nil {
-		return fmt.Errorf("%w: failed to block user", ErrInternal)
+		return fmt.Errorf("%w: failed to block user: %v", ErrInternal, err)
 	}
 
 	slog.Info("user blocked", "blocker_id", blockerID, "target_id", targetID)
@@ -59,7 +59,7 @@ func (s *BlockService) UnblockUser(ctx context.Context, blockerID, targetID int6
 		return fmt.Errorf("%w: user_id must be positive", ErrBadRequest)
 	}
 	if err := s.st.UnblockUser(ctx, blockerID, targetID); err != nil {
-		return fmt.Errorf("%w: failed to unblock user", ErrInternal)
+		return fmt.Errorf("%w: failed to unblock user: %v", ErrInternal, err)
 	}
 	slog.Info("user unblocked", "blocker_id", blockerID, "target_id", targetID)
 	return nil
@@ -69,7 +69,7 @@ func (s *BlockService) UnblockUser(ctx context.Context, blockerID, targetID int6
 func (s *BlockService) ListBlocked(ctx context.Context, blockerID int64) ([]int64, error) {
 	ids, err := s.st.ListBlockedUsers(ctx, blockerID)
 	if err != nil {
-		return nil, fmt.Errorf("%w: failed to list blocked users", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to list blocked users: %v", ErrInternal, err)
 	}
 	if ids == nil {
 		ids = []int64{}

Server/service/dm.go

@@ -55,7 +55,7 @@ func (s *DMService) CreateDM(ctx context.Context, userID, recipientID int64) (*C
 
 	blocked, err := s.st.IsEitherBlocked(ctx, userID, recipientID)
 	if err != nil {
-		return nil, fmt.Errorf("%w: failed to check block status", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to check block status: %v", ErrInternal, err)
 	}
 	if blocked {
 		return nil, fmt.Errorf("%w: cannot create DM — user is blocked", ErrForbidden)
@@ -78,7 +78,7 @@ func (s *DMService) CreateDM(ctx context.Context, userID, recipientID int64) (*C
 func (s *DMService) ListDMs(ctx context.Context, userID int64) ([]db.DMChannelInfo, error) {
 	dms, err := s.st.GetUserDMChannels(ctx, userID)
 	if err != nil {
-		return nil, fmt.Errorf("%w: failed to list DMs", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to list DMs: %v", ErrInternal, err)
 	}
 	return dms, nil
 }
@@ -95,7 +95,7 @@ func (s *DMService) CloseDM(ctx context.Context, userID, channelID int64) error
 	}
 
 	if err := s.st.CloseDM(ctx, userID, channelID); err != nil {
-		return fmt.Errorf("%w: failed to close DM", ErrInternal)
+		return fmt.Errorf("%w: failed to close DM: %v", ErrInternal, err)
 	}
 
 	slog.Debug("DM closed", "user_id", userID, "channel_id", channelID)

Server/service/invite.go

@@ -50,12 +50,12 @@ func (s *InviteService) CreateInvite(ctx context.Context, createdBy int64, maxUs
 
 	code, err := s.st.CreateInvite(ctx, createdBy, maxUses, expiresAt)
 	if err != nil {
-		return nil, fmt.Errorf("%w: failed to create invite", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to create invite: %v", ErrInternal, err)
 	}
 
 	invite, err := s.st.GetInvite(ctx, code)
 	if err != nil || invite == nil {
-		return nil, fmt.Errorf("%w: failed to retrieve invite", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to retrieve invite: %v", ErrInternal, err)
 	}
 	return invite, nil
 }
@@ -64,7 +64,7 @@ func (s *InviteService) CreateInvite(ctx context.Context, createdBy int64, maxUs
 func (s *InviteService) ListInvites(ctx context.Context) ([]*db.Invite, error) {
 	invites, err := s.st.ListInvites(ctx)
 	if err != nil {
-		return nil, fmt.Errorf("%w: failed to list invites", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to list invites: %v", ErrInternal, err)
 	}
 	return invites, nil
 }
@@ -76,7 +76,7 @@ func (s *InviteService) RevokeInvite(ctx context.Context, code string) error {
 		return fmt.Errorf("%w: invite not found", ErrNotFound)
 	}
 	if err := s.st.RevokeInvite(ctx, code); err != nil {
-		return fmt.Errorf("%w: failed to revoke invite", ErrInternal)
+		return fmt.Errorf("%w: failed to revoke invite: %v", ErrInternal, err)
 	}
 	return nil
 }

Server/service/message.go

@@ -559,7 +559,7 @@ func (s *MessageService) SearchMessages(ctx context.Context, userID int64, query
 		}
 		results, err := s.st.SearchMessages(ctx, query, channelID, limit)
 		if err != nil {
-			return nil, fmt.Errorf("%w: search failed", ErrInternal)
+			return nil, fmt.Errorf("%w: search failed: %v", ErrInternal, err)
 		}
 		return results, nil
 	}
@@ -575,7 +575,7 @@ func (s *MessageService) SearchMessages(ctx context.Context, userID int64, query
 
 	results, err := s.st.SearchMessagesInChannels(ctx, query, accessibleIDs, limit)
 	if err != nil {
-		return nil, fmt.Errorf("%w: search failed", ErrInternal)
+		return nil, fmt.Errorf("%w: search failed: %v", ErrInternal, err)
 	}
 	return results, nil
 }
@@ -599,7 +599,7 @@ func (s *MessageService) GetPinnedMessages(ctx context.Context, userID, channelI
 	}
 	msgs, err := s.st.GetPinnedMessages(ctx, channelID, userID)
 	if err != nil {
-		return nil, fmt.Errorf("%w: failed to fetch pinned messages", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to fetch pinned messages: %v", ErrInternal, err)
 	}
 	return msgs, nil
 }
@@ -633,12 +633,12 @@ func (s *MessageService) SetMessagePinned(ctx context.Context, userID, channelID
 func (s *MessageService) GetAccessibleChannelIDs(ctx context.Context, userID int64) ([]int64, error) {
 	channels, err := s.st.ListChannels(ctx)
 	if err != nil {
-		return nil, fmt.Errorf("%w: failed to list channels", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to list channels: %v", ErrInternal, err)
 	}
 
 	role, err := s.perms.GetRoleForUser(ctx, userID)
 	if err != nil || role == nil {
-		return nil, fmt.Errorf("%w: failed to get role", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to get role: %v", ErrInternal, err)
 	}
 
 	var overrides map[int64]db.ChannelOverride
@@ -646,7 +646,7 @@ func (s *MessageService) GetAccessibleChannelIDs(ctx context.Context, userID int
 		var overrideErr error
 		overrides, overrideErr = s.st.GetAllChannelPermissionsForRole(ctx, role.ID)
 		if overrideErr != nil {
-			return nil, fmt.Errorf("%w: failed to fetch channel overrides", ErrInternal)
+			return nil, fmt.Errorf("%w: failed to fetch channel overrides: %v", ErrInternal, overrideErr)
 		}
 	}
 
@@ -693,7 +693,7 @@ func (s *MessageService) checkSendPermission(ctx context.Context, userID, channe
 	if isDM {
 		ok, err := s.st.IsDMParticipant(ctx, userID, channelID)
 		if err != nil {
-			return fmt.Errorf("%w: failed to check DM participation", ErrInternal)
+			return fmt.Errorf("%w: failed to check DM participation: %v", ErrInternal, err)
 		}
 		if !ok {
 			return fmt.Errorf("%w: not a participant in this DM", ErrForbidden)
@@ -702,7 +702,7 @@ func (s *MessageService) checkSendPermission(ctx context.Context, userID, channe
 		if err == nil && recipient != nil {
 			blocked, blkErr := s.st.IsEitherBlocked(ctx, userID, recipient.ID)
 			if blkErr != nil {
-				return fmt.Errorf("%w: failed to check block status", ErrInternal)
+				return fmt.Errorf("%w: failed to check block status: %v", ErrInternal, blkErr)
 			}
 			if blocked {
 				return fmt.Errorf("%w: cannot send messages — user is blocked", ErrBlocked)

Server/service/moderation.go

@@ -96,7 +96,7 @@ func (s *ModerationService) BanUser(ctx context.Context, actorID, targetID int64
 	}
 
 	if err := s.st.BanUser(ctx, targetID, reason, expires); err != nil {
-		return fmt.Errorf("%w: failed to ban user", ErrInternal)
+		return fmt.Errorf("%w: failed to ban user: %v", ErrInternal, err)
 	}
 
 	// Audit rows must survive a request canceled after the ban committed.
@@ -125,7 +125,7 @@ func (s *ModerationService) UnbanUser(ctx context.Context, actorID, targetID int
 	}
 
 	if err := s.st.UnbanUser(ctx, targetID); err != nil {
-		return fmt.Errorf("%w: failed to unban user", ErrInternal)
+		return fmt.Errorf("%w: failed to unban user: %v", ErrInternal, err)
 	}
 
 	// Audit rows must survive a request canceled after the unban committed.

Server/service/user.go

@@ -38,11 +38,11 @@ func (s *UserService) UpdateProfile(ctx context.Context, userID int64, username
 		if db.IsUniqueConstraintError(err) {
 			return nil, fmt.Errorf("%w: username is already taken", ErrConflict)
 		}
-		return nil, fmt.Errorf("%w: failed to update profile", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to update profile: %v", ErrInternal, err)
 	}
 	user, err := s.st.GetUserByID(ctx, userID)
 	if err != nil {
-		return nil, fmt.Errorf("%w: failed to fetch updated user", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to fetch updated user: %v", ErrInternal, err)
 	}
 	// Audit rows must survive a request canceled after the write committed.
 	db.WriteAudit(context.WithoutCancel(ctx), s.st, userID, "profile_update", "user", userID,
@@ -82,7 +82,7 @@ type ChangePasswordResult struct {
 // ChangePassword updates the user's password and revokes other sessions.
 func (s *UserService) ChangePassword(ctx context.Context, userID int64, newPasswordHash string, keepSessionID int64) (ChangePasswordResult, error) {
 	if err := s.st.UpdateUserPassword(ctx, userID, newPasswordHash); err != nil {
-		return ChangePasswordResult{}, fmt.Errorf("%w: failed to update password", ErrInternal)
+		return ChangePasswordResult{}, fmt.Errorf("%w: failed to update password: %v", ErrInternal, err)
 	}
 
 	// The password is committed from here on: every path below reports
@@ -114,7 +114,7 @@ func (s *UserService) ChangePassword(ctx context.Context, userID int64, newPassw
 func (s *UserService) ListSessions(ctx context.Context, userID int64) ([]db.Session, error) {
 	sessions, err := s.st.ListUserSessions(ctx, userID)
 	if err != nil {
-		return nil, fmt.Errorf("%w: failed to list sessions", ErrInternal)
+		return nil, fmt.Errorf("%w: failed to list sessions: %v", ErrInternal, err)
 	}
 	return sessions, nil
 }
@@ -125,7 +125,7 @@ func (s *UserService) RevokeSession(ctx context.Context, userID, sessionID int64
 		if errors.Is(err, db.ErrNotFound) {
 			return fmt.Errorf("%w: session not found", ErrNotFound)
 		}
-		return fmt.Errorf("%w: failed to revoke session", ErrInternal)
+		return fmt.Errorf("%w: failed to revoke session: %v", ErrInternal, err)
 	}
 	// Audit rows must survive a request canceled after the delete committed.
 	db.WriteAudit(context.WithoutCancel(ctx), s.st, userID, "session_revoke", "session", sessionID, "session revoked")

Server/stackutil/stackutil.go

@@ -0,0 +1,45 @@
+// Package stackutil captures goroutine stack traces for panic logging in a
+// form that is safe to persist and stream.
+//
+// runtime.Stack embeds function argument values as raw hex words. For
+// arguments passed by value (e.g. a [32]byte room key or a [16]byte IV) that
+// exposes the actual bytes — secret material such as E2EE keys, session
+// tokens, or passwords that flowed through a panicking call. Because the
+// server tees panic logs into the admin ring buffer and streams them over SSE,
+// a single panic on a crypto or auth path could leak keys to any admin viewer.
+//
+// Capture avoids this by using runtime.Callers + CallersFrames, which yields
+// only function names and source locations — never argument data.
+package stackutil
+
+import (
+	"runtime"
+	"strconv"
+	"strings"
+)
+
+// Capture returns a compact, argument-free stack trace for the calling
+// goroutine: one "func" line followed by a "\tfile:line" line per frame. It is
+// safe to log even when the panicking function held sensitive arguments.
+func Capture() string {
+	pcs := make([]uintptr, 64)
+	n := runtime.Callers(2, pcs) // skip runtime.Callers and Capture itself
+	if n == 0 {
+		return ""
+	}
+	frames := runtime.CallersFrames(pcs[:n])
+	var b strings.Builder
+	for {
+		f, more := frames.Next()
+		b.WriteString(f.Function)
+		b.WriteString("\n\t")
+		b.WriteString(f.File)
+		b.WriteByte(':')
+		b.WriteString(strconv.Itoa(f.Line))
+		b.WriteByte('\n')
+		if !more {
+			break
+		}
+	}
+	return b.String()
+}

Server/stackutil/stackutil_test.go

@@ -0,0 +1,48 @@
+package stackutil
+
+import (
+	"strings"
+	"testing"
+)
+
+//go:noinline
+func panicWithSecretArgs(key [32]byte, token string) {
+	// key is passed by value, so runtime.Stack would render its bytes as hex.
+	_ = key
+	_ = token
+	panic("boom")
+}
+
+// TestCaptureOmitsArguments is the security guard: Capture must never emit
+// argument values. runtime.Stack renders by-value args as hex words ("0x..."),
+// so asserting the output is free of hex verifies no argument bytes leaked,
+// while still carrying the panicking function name and source location.
+func TestCaptureOmitsArguments(t *testing.T) {
+	var key [32]byte
+	for i := range key {
+		key[i] = 0xAB
+	}
+
+	var stack string
+	func() {
+		defer func() {
+			if recover() != nil {
+				stack = Capture()
+			}
+		}()
+		panicWithSecretArgs(key, "super-secret-token")
+	}()
+
+	if stack == "" {
+		t.Fatal("expected a captured stack after recover")
+	}
+	if strings.Contains(stack, "0x") {
+		t.Errorf("captured stack contains hex argument words (possible secret leak):\n%s", stack)
+	}
+	if !strings.Contains(stack, "panicWithSecretArgs") {
+		t.Errorf("captured stack missing the panicking function name:\n%s", stack)
+	}
+	if !strings.Contains(stack, "stackutil_test.go:") {
+		t.Errorf("captured stack missing source location:\n%s", stack)
+	}
+}

Server/telemetry/telemetry_default.go

@@ -20,3 +20,7 @@ func Init(_ context.Context, cfg config.TelemetryConfig) (ShutdownFunc, error) {
 	SetGlobal(noopProvider{})
 	return func(context.Context) error { return nil }, nil
 }
+
+// TraceIDFromContext returns the active trace ID as a hex string, or "" when no
+// span is active. The default build has no tracing, so it always returns "".
+func TraceIDFromContext(_ context.Context) string { return "" }

Server/telemetry/telemetry_otel.go

@@ -178,6 +178,16 @@ func (p *otelProvider) HTTPMiddleware(next http.Handler) http.Handler {
 // exporter registry.
 func (p *otelProvider) PrometheusHandler() http.Handler { return p.promHandler }
 
+// TraceIDFromContext returns the active trace ID as a hex string, or "" when no
+// span is recording in ctx. Used to stamp log records with trace_id so logs
+// correlate with traces.
+func TraceIDFromContext(ctx context.Context) string {
+	if sc := trace.SpanContextFromContext(ctx); sc.HasTraceID() {
+		return sc.TraceID().String()
+	}
+	return ""
+}
+
 // ── Tracer / Span adapters ─────────────────────────────────────────────────
 
 type otelTracer struct{ inner trace.Tracer }

Server/ws/hub.go

@@ -6,7 +6,6 @@ import (
 	"context"
 	"fmt"
 	"log/slog"
-	"runtime"
 	"sync"
 	"sync/atomic"
 	"time"
@@ -16,6 +15,7 @@ import (
 	"github.com/owncord/server/permissions"
 	"github.com/owncord/server/plugin"
 	"github.com/owncord/server/service"
+	"github.com/owncord/server/stackutil"
 	"github.com/owncord/server/syncutil"
 )
 
@@ -276,12 +276,10 @@ func (h *Hub) Run() {
 					}
 					panicCount++
 
-					buf := make([]byte, 4096)
-					n := runtime.Stack(buf, false)
 					slog.Error("hub: panic recovered",
 						"panic", r,
 						"panic_count", panicCount,
-						"stack", string(buf[:n]))
+						"stack", stackutil.Capture())
 
 					if panicCount >= 3 {
 						slog.Error("hub: too many panics in 60s, stopping")

Server/ws/registry.go

@@ -4,7 +4,8 @@ import (
 	"context"
 	"fmt"
 	"log/slog"
-	"runtime"
+
+	"github.com/owncord/server/stackutil"
 )
 
 // handlerV2Entry pairs a V2 handler with its domain-specific dependency struct.
@@ -47,15 +48,13 @@ func (r *HandlerRegistry) DispatchV2(ctx context.Context, cmd Command, info Clie
 	}
 	defer func() {
 		if rec := recover(); rec != nil {
-			buf := make([]byte, 4096)
-			n := runtime.Stack(buf, false)
-			// TODO: stack trace may contain sensitive function arguments
-			// (e.g. encrypted keys). Consider scrubbing or limiting frames.
+			// stackutil.Capture omits argument values, which for E2EE
+			// handlers can include encrypted key material.
 			slog.Error("DispatchV2 panic recovered",
 				"type", cmd.Type(),
 				"user_id", info.UserID,
 				"panic", rec,
-				"stack", string(buf[:n]),
+				"stack", stackutil.Capture(),
 			)
 			result = Result{Error: ClientError{Code: ErrCodeInternal, Message: "internal error"}}
 			ok = true

Server/ws/serve.go

@@ -503,6 +503,11 @@ func authenticateConn(parent context.Context, conn *websocket.Conn, database *db
 	sess, err := database.GetSessionByTokenHash(ctx, hash)
 	if err != nil || sess == nil {
 		_ = conn.Write(ctx, websocket.MessageText, buildAuthError("invalid token"))
+		if err != nil {
+			// DB outage, not a bad token — carry the cause so the caller's log
+			// distinguishes it from an ordinary invalid-token rejection.
+			return nil, "", 0, fmt.Errorf("auth: session lookup failed: %w", err)
+		}
 		return nil, "", 0, fmt.Errorf("auth: invalid session")
 	}
 
@@ -514,6 +519,9 @@ func authenticateConn(parent context.Context, conn *websocket.Conn, database *db
 	user, err := database.GetUserByID(ctx, sess.UserID)
 	if err != nil || user == nil {
 		_ = conn.Write(ctx, websocket.MessageText, buildAuthError("user not found"))
+		if err != nil {
+			return nil, "", 0, fmt.Errorf("auth: user lookup failed: %w", err)
+		}
 		return nil, "", 0, fmt.Errorf("auth: user not found")
 	}
 

Server/ws/voice_e2ee.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"encoding/base64"
 	"fmt"
+	"log/slog"
 	"time"
 )
 
@@ -226,10 +227,18 @@ func (h *Hub) sendToUserIfInVoiceChannel(voiceChannelID, targetUserID int64, msg
 
 	target, ok := h.clients[targetUserID]
 	if !ok {
-		return // target not connected — silently drop
+		// Undeliverable key offer can leave the peer unable to decrypt — the
+		// payload is dropped, but log (IDs only, never the encrypted key) so
+		// the failure is diagnosable rather than silent.
+		slog.Debug("e2ee: key offer dropped, target not connected",
+			"target_user_id", targetUserID, "voice_channel_id", voiceChannelID)
+		return
 	}
 	if target.getVoiceChID() != voiceChannelID {
-		return // target not in expected voice channel — silently drop
+		slog.Debug("e2ee: key offer dropped, target not in expected voice channel",
+			"target_user_id", targetUserID, "voice_channel_id", voiceChannelID,
+			"target_voice_channel_id", target.getVoiceChID())
+		return
 	}
 	target.sendMsg(msg)
 }