docs: explain the two DoT cert modes in README

Expands the DoT paragraph to make the trust model explicit. The
previous version said "self-signed or bring your own cert" without
explaining when to pick which or what the user experience looks like.

The two modes close numa's gap vs AdGuard Home: BYO cert mode is
functionally identical (Let's Encrypt via DNS-01 + cert_path/key_path),
and the self-signed mode is numa's advantage on LAN-only deploys.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

.github/workflows/ci.yml

@@ -27,6 +27,17 @@ jobs:
       - name: audit
         run: cargo install cargo-audit && cargo audit
 
+  check-macos:
+    runs-on: macos-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: dtolnay/rust-toolchain@stable
+      - uses: Swatinem/rust-cache@v2
+      - name: clippy
+        run: cargo clippy -- -D warnings
+      - name: test
+        run: cargo test
+
   check-windows:
     runs-on: windows-latest
     steps:
@@ -37,3 +48,10 @@ jobs:
         run: cargo build
       - name: clippy
         run: cargo clippy -- -D warnings
+      - name: test
+        run: cargo test
+      - name: Upload binary
+        uses: actions/upload-artifact@v4
+        with:
+          name: numa-windows-x86_64
+          path: target/debug/numa.exe

Cargo.lock

@@ -1143,7 +1143,7 @@ dependencies = [
 
 [[package]]
 name = "numa"
-version = "0.7.1"
+version = "0.10.0"
 dependencies = [
  "arc-swap",
  "axum",
@@ -1159,6 +1159,7 @@ dependencies = [
  "reqwest",
  "ring",
  "rustls",
+ "rustls-pemfile",
  "serde",
  "serde_json",
  "socket2 0.5.10",
@@ -1546,6 +1547,15 @@ dependencies = [
  "zeroize",
 ]
 
+[[package]]
+name = "rustls-pemfile"
+version = "2.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "dce314e5fee3f39953d46bb63bb8a46d40c2f8fb7cc5a3b6cab2bde9721d6e50"
+dependencies = [
+ "rustls-pki-types",
+]
+
 [[package]]
 name = "rustls-pki-types"
 version = "1.14.0"

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "numa"
-version = "0.7.1"
+version = "0.10.0"
 authors = ["razvandimescu <razvan@dimescu.com>"]
 edition = "2021"
 description = "Portable DNS resolver in Rust — .numa local domains, ad blocking, developer overrides, DNS-over-HTTPS"
@@ -29,6 +29,7 @@ rustls = "0.23"
 tokio-rustls = "0.26"
 arc-swap = "1"
 ring = "0.17"
+rustls-pemfile = "2.2.0"
 
 [dev-dependencies]
 criterion = { version = "0.5", features = ["html_reports"] }

Dockerfile

@@ -13,5 +13,5 @@ RUN cargo build --release
 
 FROM alpine:3.20
 COPY --from=builder /app/target/release/numa /usr/local/bin/numa
-EXPOSE 53/udp 80/tcp 443/tcp 5380/tcp
+EXPOSE 53/udp 80/tcp 443/tcp 853/tcp 5380/tcp
 ENTRYPOINT ["numa"]

README.md

@@ -8,189 +8,127 @@
 
 A portable DNS resolver in a single binary. Block ads on any network, name your local services (`frontend.numa`), and override any hostname with auto-revert — all from your laptop, no cloud account or Raspberry Pi required.
 
-Built from scratch in Rust. Zero DNS libraries. RFC 1035 wire protocol parsed by hand. Recursive resolution from root nameservers with full DNSSEC validation (chain-of-trust + NSEC/NSEC3 denial proofs). One ~8MB binary, no PHP, no web server, no database — everything is embedded.
+Built from scratch in Rust. Zero DNS libraries. RFC 1035 wire protocol parsed by hand. Caching, ad blocking, and local service domains out of the box. Optional recursive resolution from root nameservers with full DNSSEC chain-of-trust validation, plus a DNS-over-TLS listener for encrypted client connections (iOS Private DNS, systemd-resolved, etc.). One ~8MB binary, everything embedded.
 
 ![Numa dashboard](assets/hero-demo.gif)
 
 ## Quick Start
 
 ```bash
-# Install (pick one)
+# macOS
 brew install razvandimescu/tap/numa
-cargo install numa
+
+# Linux
 curl -fsSL https://raw.githubusercontent.com/razvandimescu/numa/main/install.sh | sh
 
-# Run (port 53 requires root)
-sudo numa
+# Windows — download from GitHub Releases
+# All platforms
+cargo install numa
+```
 
-# Try it
-dig @127.0.0.1 google.com           # ✓ resolves normally
-dig @127.0.0.1 ads.google.com       # ✗ blocked → 0.0.0.0
+```bash
+sudo numa                              # run in foreground (port 53 requires root/admin)
 ```
 
 Open the dashboard: **http://numa.numa** (or `http://localhost:5380`)
 
-### Set as system resolver
+Set as system DNS:
 
-```bash
-# Point your system DNS to Numa (saves originals for uninstall)
-sudo numa install
+| Platform | Install | Uninstall |
+|----------|---------|-----------|
+| macOS | `sudo numa install` | `sudo numa uninstall` |
+| Linux | `sudo numa install` | `sudo numa uninstall` |
+| Windows | `numa install` (admin) + reboot | `numa uninstall` (admin) + reboot |
 
-# Run as a persistent service (auto-starts on boot, restarts if killed)
-sudo numa service start
-```
+On macOS and Linux, numa runs as a system service (launchd/systemd). On Windows, numa auto-starts on login via registry.
 
-To uninstall: `sudo numa service stop` removes the service, `sudo numa uninstall` restores your original DNS.
+## Local Services
 
-### Upgrade
-
-```bash
-# From Homebrew
-brew upgrade numa
-
-# From source
-make deploy    # builds release, copies binary, re-signs, restarts service
-```
-
-### Build from source
-
-```bash
-git clone https://github.com/razvandimescu/numa.git && cd numa
-cargo build --release
-sudo cp target/release/numa /usr/local/bin/numa
-```
-
-## Why Numa
-
-- **Local service proxy** — `https://frontend.numa` instead of `localhost:5173`. Auto-generated TLS certs, WebSocket support for HMR. Like `/etc/hosts` but with auto TLS, a REST API, LAN discovery, and auto-revert.
-- **Path-based routing** — `app.numa/api → :5001`, `app.numa/auth → :5002`. Route URL paths to different backends with optional prefix stripping. Like nginx location blocks, zero config files.
-- **LAN service discovery** — Numa instances on the same network find each other automatically via mDNS. Access a teammate's `api.numa` from your machine. Opt-in via `[lan] enabled = true`.
-- **Developer overrides** — point any hostname to any IP, auto-reverts after N minutes. Full REST API for scripting. Built-in diagnostics: `curl localhost:5380/diagnose/example.com` tells you exactly how any domain resolves.
-- **DNS-over-HTTPS** — upstream queries encrypted via DoH. Your ISP sees HTTPS traffic, not DNS queries. Set `address = "https://9.9.9.9/dns-query"` in `[upstream]` or any DoH provider.
-- **Ad blocking that travels with you** — 385K+ domains blocked via [Hagezi Pro](https://github.com/hagezi/dns-blocklists). Works on any network: coffee shops, hotels, airports.
-- **Sub-microsecond caching** — 691ns cached round-trip, ~2.0M queries/sec throughput, zero heap allocations in the I/O path. [Benchmarks](bench/).
-- **Live dashboard** — real-time stats, query log, blocking controls, service management. LAN accessibility badges show which services are reachable from other devices.
-- **macOS, Linux, and Windows** — `numa install` configures system DNS, `numa service start` runs as launchd/systemd service.
-
-## Local Service Proxy
-
-Name your local dev services with `.numa` domains:
+Name your dev services instead of remembering port numbers:
 
 ```bash
 curl -X POST localhost:5380/services \
-  -H 'Content-Type: application/json' \
   -d '{"name":"frontend","target_port":5173}'
-
-open http://frontend.numa            # → proxied to localhost:5173
 ```
 
-- **HTTPS with green lock** — auto-generated local CA + per-service TLS certs
-- **WebSocket** — Vite/webpack HMR works through the proxy
-- **Health checks** — dashboard shows green/red status per service
-- **LAN sharing** — services bound to `0.0.0.0` are automatically discoverable by other Numa instances on the network. Dashboard shows "LAN" or "local only" per service.
-- **Path-based routing** — route URL paths to different backends:
-  ```toml
-  [[services]]
-  name = "app"
-  target_port = 3000
-  routes = [
-      { path = "/api", port = 5001 },
-      { path = "/auth", port = 5002, strip = true },
-  ]
-  ```
-  `app.numa/api/users → :5001/api/users`, `app.numa/auth/login → :5002/login` (stripped)
-- **Persistent** — services survive restarts
-- Or configure in `numa.toml`:
+Now `https://frontend.numa` works in your browser — green lock, valid cert, WebSocket passthrough for HMR. No mkcert, no nginx, no `/etc/hosts`.
 
-```toml
-[[services]]
-name = "frontend"
-target_port = 5173
-```
+Add path-based routing (`app.numa/api → :5001`), share services across machines via LAN discovery, or configure everything in [`numa.toml`](numa.toml).
 
-## LAN Service Discovery
+## Ad Blocking & Privacy
 
-Run Numa on multiple machines. They find each other automatically:
+385K+ domains blocked via [Hagezi Pro](https://github.com/hagezi/dns-blocklists). Works on any network — coffee shops, hotels, airports. Travels with your laptop.
+
+Three resolution modes:
+
+- **`forward`** (default) — transparent proxy to your existing system DNS. Everything works as before, just with caching and ad blocking on top. Captive portals, VPNs, corporate DNS — all respected.
+- **`recursive`** — resolve directly from root nameservers. No upstream dependency, no single entity sees your full query pattern. Add `[dnssec] enabled = true` for full chain-of-trust validation.
+- **`auto`** — probe root servers on startup, recursive if reachable, encrypted DoH fallback if blocked.
+
+DNSSEC validates the full chain of trust: RRSIG signatures, DNSKEY verification, DS delegation, NSEC/NSEC3 denial proofs. [Read how it works →](https://numa.rs/blog/posts/dnssec-from-scratch.html)
+
+**DNS-over-TLS listener** (RFC 7858) — accept encrypted queries on port 853 from strict clients like iOS Private DNS, systemd-resolved, or stubby. Two modes:
+
+- **Self-signed** (default) — numa generates a local CA automatically. Works on any network with zero DNS setup, but clients must manually trust the CA (on macOS/Linux add to the system trust store; on iOS install a `.mobileconfig`).
+- **Bring-your-own cert** — point `[dot] cert_path` / `key_path` at a publicly-trusted cert (e.g., Let's Encrypt via DNS-01 challenge on a domain pointing at your numa instance). Clients connect without any trust-store setup — same UX as AdGuard Home or Cloudflare `1.1.1.1`.
+
+ALPN `"dot"` is advertised and enforced in both modes; a handshake with mismatched ALPN is rejected as a cross-protocol confusion defense.
+
+## LAN Discovery
+
+Run Numa on multiple machines. They find each other automatically via mDNS:
 
 ```
 Machine A (192.168.1.5)              Machine B (192.168.1.20)
 ┌──────────────────────┐             ┌──────────────────────┐
 │ Numa                 │    mDNS     │ Numa                 │
-│  services:           │◄───────────►│  services:           │
-│   - api (port 8000)  │  discovery  │   - grafana (3000)   │
-│   - frontend (5173)  │             │                      │
+│  - api (port 8000)   │◄───────────►│  - grafana (3000)    │
+│  - frontend (5173)   │  discovery  │                      │
 └──────────────────────┘             └──────────────────────┘
 ```
 
-From Machine B:
-```bash
-dig @127.0.0.1 api.numa          # → 192.168.1.5
-curl http://api.numa              # → proxied to Machine A's port 8000
-```
+From Machine B: `curl http://api.numa` → proxied to Machine A's port 8000. Enable with `numa lan on`.
 
-Enable LAN discovery:
-```bash
-numa lan on
-```
-Or in `numa.toml`:
-```toml
-[lan]
-enabled = true
-```
-Uses standard mDNS (`_numa._tcp.local` on port 5353) — compatible with Bonjour/Avahi, silently dropped by corporate firewalls instead of triggering IPS alerts.
-
-**Hub mode** — don't want to install Numa on every machine? Run one instance as a shared DNS server and point other devices to it:
-
-```bash
-# On the hub machine, bind to LAN interface
-[server]
-bind_addr = "0.0.0.0:53"
-
-# On other devices, set DNS to the hub's IP
-# They get .numa resolution, ad blocking, caching — zero install
-```
+**Hub mode**: run one instance with `bind_addr = "0.0.0.0:53"` and point other devices' DNS to it — they get ad blocking + `.numa` resolution without installing anything.
 
 ## How It Compares
 
-| | Pi-hole | AdGuard Home | NextDNS | Cloudflare | Numa |
-|---|---|---|---|---|---|
-| Local service proxy | No | No | No | No | `.numa` + HTTPS + WS |
-| Path-based routing | No | No | No | No | Prefix match + strip |
-| LAN service discovery | No | No | No | No | mDNS, opt-in |
-| Developer overrides | No | No | No | No | REST API + auto-expiry |
-| Recursive resolver | No | No | Cloud only | Cloud only | From root hints, DNSSEC |
-| Encrypted upstream (DoH) | No (needs cloudflared) | Yes | Cloud only | Cloud only | Native, single binary |
-| Portable (travels with laptop) | No (appliance) | No (appliance) | Cloud only | Cloud only | Single binary |
-| Zero config | Complex | Docker/setup | Yes | Yes | Works out of the box |
-| Ad blocking | Yes | Yes | Yes | Limited | 385K+ domains |
-| Data stays local | Yes | Yes | Cloud | Cloud | 100% local |
+| | Pi-hole | AdGuard Home | Unbound | Numa |
+|---|---|---|---|---|
+| Local service proxy + auto TLS | — | — | — | `.numa` domains, HTTPS, WebSocket |
+| LAN service discovery | — | — | — | mDNS, zero config |
+| Developer overrides (REST API) | — | — | — | Auto-revert, scriptable |
+| Recursive resolver | — | — | Yes | Yes, with SRTT selection |
+| DNSSEC validation | — | — | Yes | Yes (RSA, ECDSA, Ed25519) |
+| Ad blocking | Yes | Yes | — | 385K+ domains |
+| Web admin UI | Full | Full | — | Dashboard |
+| Encrypted upstream (DoH) | Needs cloudflared | Yes | — | Native |
+| Encrypted clients (DoT listener) | Needs stunnel sidecar | Yes | Yes | Native (RFC 7858) |
+| Portable (laptop) | No (appliance) | No (appliance) | Server | Single binary, macOS/Linux/Windows |
+| Community maturity | 56K stars, 10 years | 33K stars | 20 years | New |
 
-## How It Works
+## Performance
 
-```
-Query → Overrides → .numa TLD → Blocklist → Local Zones → Cache → Recursive/Forward
-```
+691ns cached round-trip. ~2.0M qps throughput. Zero heap allocations in the hot path. Recursive queries average 237ms after SRTT warmup (12x improvement over round-robin). ECDSA P-256 DNSSEC verification: 174ns. [Benchmarks →](bench/)
 
-Two resolution modes: **forward** (relay to upstream like Quad9/Cloudflare) or **recursive** (resolve from root nameservers — no upstream dependency). Set `mode = "recursive"` in `[upstream]` to resolve independently.
+## Learn More
 
-No DNS libraries — no `hickory-dns`, no `trust-dns`. The wire protocol — headers, labels, compression pointers, record types — is parsed and serialized by hand. Runs on `tokio` + `axum`, async per-query task spawning.
-
-[Configuration reference](numa.toml)
+- [Blog: Implementing DNSSEC from Scratch in Rust](https://numa.rs/blog/posts/dnssec-from-scratch.html)
+- [Blog: I Built a DNS Resolver from Scratch](https://numa.rs/blog/posts/dns-from-scratch.html)
+- [Configuration reference](numa.toml) — all options documented inline
+- [REST API](src/api.rs) — 27 endpoints across overrides, cache, blocking, services, diagnostics
 
 ## Roadmap
 
-- [x] DNS proxy core — forwarding, caching, local zones
-- [x] Developer overrides — REST API with auto-expiry
-- [x] Ad blocking — 385K+ domains, live dashboard, allowlist
-- [x] System integration — macOS + Linux, launchd/systemd, Tailscale/VPN auto-discovery
-- [x] Local service proxy — `.numa` domains, HTTP/HTTPS proxy, auto TLS, WebSocket
-- [x] Path-based routing — URL prefix routing with optional strip, REST API
-- [x] LAN service discovery — mDNS auto-discovery (opt-in), cross-machine DNS + proxy
-- [x] DNS-over-HTTPS — encrypted upstream via DoH (Quad9, Cloudflare, any provider)
-- [x] Recursive resolution — resolve from root nameservers, no upstream dependency
-- [x] DNSSEC validation — chain-of-trust, NSEC/NSEC3 denial proofs, AD bit (RSA, ECDSA, Ed25519)
-- [ ] pkarr integration — self-sovereign DNS via Mainline DHT (15M nodes)
-- [ ] Global `.numa` names — self-publish, DHT-backed, first-come-first-served
+- [x] DNS forwarding, caching, ad blocking, developer overrides
+- [x] `.numa` local domains — auto TLS, path routing, WebSocket proxy
+- [x] LAN service discovery — mDNS, cross-machine DNS + proxy
+- [x] DNS-over-HTTPS — encrypted upstream
+- [x] DNS-over-TLS listener — encrypted client connections (RFC 7858, ALPN strict)
+- [x] Recursive resolution + DNSSEC — chain-of-trust, NSEC/NSEC3
+- [x] SRTT-based nameserver selection
+- [ ] pkarr integration — self-sovereign DNS via Mainline DHT
+- [ ] Global `.numa` names — DHT-backed, no registrar
 
 ## License
 

com.numa.dns.plist

@@ -6,7 +6,7 @@
     <string>com.numa.dns</string>
     <key>ProgramArguments</key>
     <array>
-        <string>/usr/local/bin/numa</string>
+        <string>{{exe_path}}</string>
     </array>
     <key>RunAtLoad</key>
     <true/>

install.sh

@@ -70,8 +70,10 @@ echo ""
 echo "  \033[38;2;107;124;78mInstalled:\033[0m $INSTALL_DIR/numa ($TAG)"
 echo ""
 echo "  Get started:"
-echo "    sudo numa                    # start the DNS server"
-echo "    sudo numa install            # set as system DNS"
-echo "    sudo numa service start      # run as persistent service"
-echo "    open http://localhost:5380    # dashboard"
+echo "    sudo numa install            # install service + set as system DNS"
+echo "    open http://localhost:5380   # dashboard"
+echo ""
+echo "  Other commands:"
+echo "    sudo numa                    # run in foreground (no service)"
+echo "    sudo numa uninstall          # restore original DNS"
 echo ""

numa.service

@@ -5,7 +5,7 @@ Wants=network-online.target
 
 [Service]
 Type=simple
-ExecStart=/usr/local/bin/numa
+ExecStart={{exe_path}}
 Restart=always
 RestartSec=2
 StandardOutput=journal

numa.toml

@@ -2,6 +2,11 @@
 bind_addr = "0.0.0.0:53"
 api_port = 5380
 # api_bind_addr = "127.0.0.1"  # default; set to "0.0.0.0" for LAN dashboard access
+# data_dir = "/usr/local/var/numa"  # where numa stores TLS CA and cert material
+                                    # (default: /usr/local/var/numa on unix,
+                                    # %PROGRAMDATA%\numa on windows). Override for
+                                    # containerized deploys or tests that can't
+                                    # write to the system path.
 
 # [upstream]
 # mode = "forward"                              # "forward" (default) — relay to upstream
@@ -54,7 +59,7 @@ enabled = true
 port = 80
 tls_port = 443
 tld = "numa"
-# bind_addr = "127.0.0.1"  # default; auto 0.0.0.0 when [lan] enabled
+# bind_addr = "127.0.0.1"  # default; set to "0.0.0.0" for LAN access to .numa services
 
 # Pre-configured services (numa.numa is always added automatically)
 # [[services]]
@@ -83,6 +88,14 @@ tld = "numa"
 # enabled = false             # opt-in: verify chain of trust from root KSK
 # strict = false              # true = SERVFAIL on bogus signatures
 
+# DNS-over-TLS listener (RFC 7858) — encrypted DNS on port 853
+# [dot]
+# enabled = false             # opt-in: accept DoT queries
+# port = 853                  # standard DoT port
+# bind_addr = "0.0.0.0"       # IPv4 or IPv6; unspecified binds all interfaces
+# cert_path = "/etc/numa/dot.crt"  # PEM cert; omit to use self-signed (proxy CA if available)
+# key_path = "/etc/numa/dot.key"   # PEM private key; must be set together with cert_path
+
 # LAN service discovery via mDNS (disabled by default — no network traffic unless enabled)
 # [lan]
 # enabled = true              # discover other Numa instances via mDNS (_numa._tcp.local)

site/dashboard.html

@@ -101,7 +101,7 @@ body {
 /* Stat cards row */
 .stats-row {
   display: grid;
-  grid-template-columns: repeat(5, 1fr);
+  grid-template-columns: repeat(6, 1fr);
   gap: 1rem;
 }
 .stat-card {
@@ -125,6 +125,8 @@ body {
 .stat-card.blocked::before { background: var(--rose); }
 .stat-card.overrides::before { background: var(--violet); }
 .stat-card.uptime::before { background: var(--cyan); }
+.stat-card.memory::before { background: var(--text-dim); }
+.stat-card.memory .stat-value { color: var(--text-secondary); }
 
 .stat-label {
   font-size: 0.7rem;
@@ -468,10 +470,74 @@ body {
   display: none;
 }
 
+/* Memory sidebar panel */
+.memory-bar {
+  display: flex;
+  height: 18px;
+  border-radius: 4px;
+  overflow: hidden;
+  background: var(--bg-surface);
+  margin-bottom: 0.8rem;
+}
+.memory-bar-seg {
+  height: 100%;
+  min-width: 2px;
+  transition: width 0.6s ease;
+}
+.memory-bar-seg.cache { background: var(--teal); }
+.memory-bar-seg.blocklist { background: var(--rose); }
+.memory-bar-seg.querylog { background: var(--amber); }
+.memory-bar-seg.srtt { background: var(--cyan); }
+.memory-bar-seg.overrides { background: var(--violet); }
+.memory-row {
+  display: flex;
+  align-items: center;
+  padding: 0.3rem 0;
+  border-bottom: 1px solid var(--border);
+  font-family: var(--font-mono);
+  font-size: 0.72rem;
+}
+.memory-row:last-child { border-bottom: none; }
+.memory-row-dot {
+  width: 8px;
+  height: 8px;
+  border-radius: 2px;
+  flex-shrink: 0;
+  margin-right: 0.5rem;
+}
+.memory-row-label {
+  flex: 1;
+  color: var(--text-secondary);
+}
+.memory-row-size {
+  width: 65px;
+  text-align: right;
+  color: var(--text-primary);
+  font-weight: 500;
+}
+.memory-row-entries {
+  width: 90px;
+  text-align: right;
+  color: var(--text-dim);
+}
+.memory-rss {
+  margin-top: 0.5rem;
+  padding-top: 0.5rem;
+  border-top: 1px solid var(--border);
+  display: flex;
+  justify-content: space-between;
+  font-family: var(--font-mono);
+  font-size: 0.72rem;
+  color: var(--text-dim);
+}
+
 /* Responsive */
 @media (max-width: 1100px) {
   .main-grid { grid-template-columns: 1fr; }
 }
+@media (max-width: 900px) {
+  .stats-row { grid-template-columns: repeat(3, 1fr); }
+}
 @media (max-width: 700px) {
   .stats-row { grid-template-columns: repeat(2, 1fr); }
   .dashboard { padding: 1rem; }
@@ -524,6 +590,11 @@ body {
       <div class="stat-value" id="uptime">—</div>
       <div class="stat-sub" id="uptimeSub">&nbsp;</div>
     </div>
+    <div class="stat-card memory">
+      <div class="stat-label">Memory</div>
+      <div class="stat-value" id="memoryRss">—</div>
+      <div class="stat-sub" id="memorySub">&nbsp;</div>
+    </div>
   </div>
 
   <!-- Resolution paths -->
@@ -648,6 +719,17 @@ body {
         </div>
       </div>
 
+      <!-- Memory breakdown -->
+      <div class="panel" id="memoryPanel">
+        <div class="panel-header">
+          <span class="panel-title">Memory</span>
+          <span class="panel-title" id="memoryTotal" style="color: var(--text-dim)"></span>
+        </div>
+        <div class="panel-body" id="memoryBody">
+          <div class="empty-state">No memory data</div>
+        </div>
+      </div>
+
       <!-- Cache entries -->
       <div class="panel">
         <div class="panel-header">
@@ -712,6 +794,69 @@ function formatRemaining(secs) {
   return `${Math.floor(secs / 3600)}h ${Math.floor((secs % 3600) / 60)}m left`;
 }
 
+function formatBytes(bytes) {
+  if (bytes === 0) return '0 B';
+  if (bytes < 1024) return bytes + ' B';
+  if (bytes < 1048576) return (bytes / 1024).toFixed(1) + ' KB';
+  if (bytes < 1073741824) return (bytes / 1048576).toFixed(1) + ' MB';
+  return (bytes / 1073741824).toFixed(1) + ' GB';
+}
+
+const MEMORY_COMPONENTS = [
+  { key: 'cache',     label: 'Cache',     cls: 'cache',     color: 'var(--teal)' },
+  { key: 'blocklist', label: 'Blocklist',  cls: 'blocklist', color: 'var(--rose)' },
+  { key: 'query_log', label: 'Query Log',  cls: 'querylog',  color: 'var(--amber)' },
+  { key: 'srtt',      label: 'SRTT',       cls: 'srtt',      color: 'var(--cyan)' },
+  { key: 'overrides', label: 'Overrides',  cls: 'overrides', color: 'var(--violet)' },
+];
+
+function renderMemory(mem, stats) {
+  if (!mem) return;
+
+  // Stat card
+  document.getElementById('memoryRss').textContent = formatBytes(mem.process_memory_bytes);
+  document.getElementById('memorySub').textContent = 'est. ' + formatBytes(mem.total_estimated_bytes);
+
+  const entryCounts = {
+    cache: stats.cache.entries,
+    blocklist: stats.blocking.domains_loaded,
+    query_log: mem.query_log_entries,
+    srtt: mem.srtt_entries,
+    overrides: stats.overrides.active,
+  };
+
+  // Sidebar panel
+  const total = mem.total_estimated_bytes || 1;
+  document.getElementById('memoryTotal').textContent = formatBytes(total);
+
+  const barSegments = MEMORY_COMPONENTS.map(c => {
+    const bytes = mem[c.key + '_bytes'] || 0;
+    const pct = ((bytes / total) * 100).toFixed(1);
+    return `<div class="memory-bar-seg ${c.cls}" style="width:${pct}%" title="${c.label}: ${formatBytes(bytes)} (${pct}%)"></div>`;
+  }).join('');
+
+  const rows = MEMORY_COMPONENTS.map(c => {
+    const bytes = mem[c.key + '_bytes'] || 0;
+    const entries = entryCounts[c.key] || 0;
+    return `
+      <div class="memory-row">
+        <div class="memory-row-dot" style="background:${c.color}"></div>
+        <span class="memory-row-label">${c.label}</span>
+        <span class="memory-row-size">${formatBytes(bytes)}</span>
+        <span class="memory-row-entries">${formatNumber(entries)} entries</span>
+      </div>`;
+  }).join('');
+
+  document.getElementById('memoryBody').innerHTML = `
+    <div class="memory-bar">${barSegments}</div>
+    ${rows}
+    <div class="memory-rss">
+      <span>Process Footprint</span>
+      <span>${formatBytes(mem.process_memory_bytes)}</span>
+    </div>
+  `;
+}
+
 const PATH_DEFS = [
   { key: 'forwarded', label: 'Forward', cls: 'forward' },
   { key: 'recursive', label: 'Recursive', cls: 'recursive' },
@@ -882,6 +1027,9 @@ async function refresh() {
     document.getElementById('footerUpstream').textContent = stats.upstream || '';
     document.getElementById('footerConfig').textContent = stats.config_path || '';
     document.getElementById('footerData').textContent = stats.data_dir || '';
+    const modeEl = document.getElementById('footerMode');
+    modeEl.textContent = stats.mode || '—';
+    modeEl.style.color = stats.mode === 'recursive' ? 'var(--emerald)' : 'var(--amber)';
     document.getElementById('footerDnssec').textContent = stats.dnssec ? 'on' : 'off';
     document.getElementById('footerDnssec').style.color = stats.dnssec ? 'var(--emerald)' : 'var(--text-dim)';
     document.getElementById('footerSrtt').textContent = stats.srtt ? 'on' : 'off';
@@ -945,7 +1093,7 @@ async function refresh() {
     prevTime = now;
 
     // Cache hit rate
-    const answered = q.cached + q.forwarded + q.local + q.overridden;
+    const answered = q.cached + q.forwarded + q.recursive + q.coalesced + q.local + q.overridden;
     const hitRate = answered > 0 ? ((q.cached / answered) * 100).toFixed(1) : '0.0';
     document.getElementById('cacheRate').textContent = hitRate + '%';
 
@@ -957,6 +1105,7 @@ async function refresh() {
     renderServices(services);
     renderBlockingInfo(blockingInfo);
     renderAllowlist(allowlist);
+    renderMemory(stats.memory, stats);
 
   } catch (err) {
     document.getElementById('statusDot').className = 'status-dot error';
@@ -1236,6 +1385,7 @@ setInterval(refresh, 2000);
   Config: <span id="footerConfig" style="user-select:all;color:var(--emerald);"></span>
   · Data: <span id="footerData" style="user-select:all;color:var(--emerald);"></span>
   · Upstream: <span id="footerUpstream" style="user-select:all;color:var(--emerald);"></span>
+  · Mode: <span id="footerMode" style="color:var(--text-dim);">—</span>
   · DNSSEC: <span id="footerDnssec" style="color:var(--text-dim);">—</span>
   · SRTT: <span id="footerSrtt" style="color:var(--text-dim);">—</span>
   · Logs: <span style="user-select:all;color:var(--emerald);">macOS: /usr/local/var/log/numa.log · Linux: journalctl -u numa -f</span>

site/index.html

@@ -4,10 +4,10 @@
 <meta charset="UTF-8">
 <meta name="viewport" content="width=device-width, initial-scale=1.0">
 <title>Numa — DNS you own. Everywhere you go.</title>
-<meta name="description" content="DNS you own. Recursive resolver with full DNSSEC validation, ad blocking, .numa local domains, developer overrides. A single portable binary built from scratch in Rust.">
+<meta name="description" content="DNS you own. Portable DNS resolver with caching, ad blocking, .numa local domains, developer overrides. Optional recursive resolution with full DNSSEC validation. Built from scratch in Rust.">
 <link rel="canonical" href="https://numa.rs">
 <meta property="og:title" content="Numa — DNS you own. Everywhere you go.">
-<meta property="og:description" content="Recursive DNS resolver with full DNSSEC validation, ad blocking, .numa local domains, and developer overrides. Built from scratch in Rust.">
+<meta property="og:description" content="Portable DNS resolver with caching, ad blocking, .numa local domains, and developer overrides. Optional recursive resolution with full DNSSEC validation. Built from scratch in Rust.">
 <meta property="og:type" content="website">
 <meta property="og:url" content="https://numa.rs">
 <link rel="stylesheet" href="/fonts/fonts.css">
@@ -1232,17 +1232,17 @@ footer .closing {
     <div class="reveal">
       <div class="section-label">How It Works</div>
       <h2>What it does today</h2>
-      <p class="lead">A recursive DNS resolver with DNSSEC validation, ad blocking, local service domains, and a REST API. Everything runs in a single binary.</p>
+      <p class="lead">A DNS resolver with caching, ad blocking, local service domains, and a REST API. Optional recursive resolution with DNSSEC. Everything runs in a single binary.</p>
     </div>
     <div class="layers-grid">
       <div class="layer-card reveal reveal-delay-1">
         <div class="layer-badge">Layer 1</div>
         <h3>Resolve &amp; Protect</h3>
         <ul>
-          <li>Recursive resolution &mdash; resolve from root nameservers, no upstream needed</li>
-          <li>DNSSEC validation &mdash; chain-of-trust + NSEC/NSEC3 denial proofs (RSA, ECDSA, Ed25519)</li>
+          <li>Forward mode by default &mdash; transparent proxy to your existing DNS, with caching</li>
           <li>Ad &amp; tracker blocking &mdash; 385K+ domains, zero config</li>
-          <li>DNS-over-HTTPS &mdash; encrypted upstream as alternative to recursive mode</li>
+          <li>Recursive resolution &mdash; opt-in, resolve from root nameservers, no upstream needed</li>
+          <li>DNSSEC validation &mdash; chain-of-trust + NSEC/NSEC3 denial proofs (RSA, ECDSA, Ed25519)</li>
           <li>TTL-aware caching (sub-ms lookups)</li>
           <li>Single binary, portable &mdash; macOS, Linux, and Windows</li>
         </ul>

src/api.rs

@@ -160,6 +160,7 @@ struct QueryLogResponse {
 struct StatsResponse {
     uptime_secs: u64,
     upstream: String,
+    mode: &'static str, // "recursive" or "forward" — never "auto" at runtime
     config_path: String,
     data_dir: String,
     dnssec: bool,
@@ -169,6 +170,7 @@ struct StatsResponse {
     overrides: OverrideStats,
     blocking: BlockingStatsResponse,
     lan: LanStatsResponse,
+    memory: MemoryStats,
 }
 
 #[derive(Serialize)]
@@ -209,6 +211,19 @@ struct BlockingStatsResponse {
     allowlist_size: usize,
 }
 
+#[derive(Serialize)]
+struct MemoryStats {
+    cache_bytes: usize,
+    blocklist_bytes: usize,
+    query_log_bytes: usize,
+    query_log_entries: usize,
+    srtt_bytes: usize,
+    srtt_entries: usize,
+    overrides_bytes: usize,
+    total_estimated_bytes: usize,
+    process_memory_bytes: usize,
+}
+
 #[derive(Serialize)]
 struct DiagnoseResponse {
     domain: String,
@@ -410,14 +425,8 @@ async fn forward_query_for_diagnose(
     timeout: std::time::Duration,
 ) -> (bool, String) {
     use crate::packet::DnsPacket;
-    use crate::question::DnsQuestion;
 
-    let mut query = DnsPacket::new();
-    query.header.id = 0xBEEF;
-    query.header.recursion_desired = true;
-    query
-        .questions
-        .push(DnsQuestion::new(domain.to_string(), QueryType::A));
+    let query = DnsPacket::query(0xBEEF, domain, QueryType::A);
 
     match forward_query(&query, upstream, timeout).await {
         Ok(resp) => (
@@ -476,12 +485,29 @@ async fn query_log(
 
 async fn stats(State(ctx): State<Arc<ServerCtx>>) -> Json<StatsResponse> {
     let snap = ctx.stats.lock().unwrap().snapshot();
-    let (cache_len, cache_max) = {
+    let (cache_len, cache_max, cache_bytes) = {
         let cache = ctx.cache.read().unwrap();
-        (cache.len(), cache.max_entries())
+        (cache.len(), cache.max_entries(), cache.heap_bytes())
     };
-    let override_count = ctx.overrides.read().unwrap().active_count();
-    let bl_stats = ctx.blocklist.read().unwrap().stats();
+    let (override_count, overrides_bytes) = {
+        let ov = ctx.overrides.read().unwrap();
+        (ov.active_count(), ov.heap_bytes())
+    };
+    let (bl_stats, blocklist_bytes) = {
+        let bl = ctx.blocklist.read().unwrap();
+        (bl.stats(), bl.heap_bytes())
+    };
+    let (query_log_bytes, query_log_entries) = {
+        let log = ctx.query_log.lock().unwrap();
+        (log.heap_bytes(), log.len())
+    };
+    let (srtt_bytes, srtt_entries, srtt_enabled) = {
+        let s = ctx.srtt.read().unwrap();
+        (s.heap_bytes(), s.len(), s.is_enabled())
+    };
+
+    let total_estimated =
+        cache_bytes + blocklist_bytes + query_log_bytes + srtt_bytes + overrides_bytes;
 
     let upstream = if ctx.upstream_mode == crate::config::UpstreamMode::Recursive {
         "recursive (root hints)".to_string()
@@ -492,10 +518,11 @@ async fn stats(State(ctx): State<Arc<ServerCtx>>) -> Json<StatsResponse> {
     Json(StatsResponse {
         uptime_secs: snap.uptime_secs,
         upstream,
+        mode: ctx.upstream_mode.as_str(),
         config_path: ctx.config_path.clone(),
         data_dir: ctx.data_dir.to_string_lossy().to_string(),
         dnssec: ctx.dnssec_enabled,
-        srtt: ctx.srtt.read().unwrap().is_enabled(),
+        srtt: srtt_enabled,
         queries: QueriesStats {
             total: snap.total,
             forwarded: snap.forwarded,
@@ -524,6 +551,17 @@ async fn stats(State(ctx): State<Arc<ServerCtx>>) -> Json<StatsResponse> {
             enabled: ctx.lan_enabled,
             peers: ctx.lan_peers.lock().unwrap().list().len(),
         },
+        memory: MemoryStats {
+            cache_bytes,
+            blocklist_bytes,
+            query_log_bytes,
+            query_log_entries,
+            srtt_bytes,
+            srtt_entries,
+            overrides_bytes,
+            total_estimated_bytes: total_estimated,
+            process_memory_bytes: crate::stats::process_memory_bytes(),
+        },
     })
 }
 

src/blocklist.rs

@@ -183,6 +183,15 @@ impl BlocklistStore {
         self.allowlist.iter().cloned().collect()
     }
 
+    pub fn heap_bytes(&self) -> usize {
+        let per_slot_overhead = std::mem::size_of::<u64>() + std::mem::size_of::<String>() + 1;
+        let domains_table = self.domains.capacity() * per_slot_overhead;
+        let domains_heap: usize = self.domains.iter().map(|d| d.capacity()).sum();
+        let allow_table = self.allowlist.capacity() * per_slot_overhead;
+        let allow_heap: usize = self.allowlist.iter().map(|d| d.capacity()).sum();
+        domains_table + domains_heap + allow_table + allow_heap
+    }
+
     pub fn stats(&self) -> BlocklistStats {
         BlocklistStats {
             enabled: self.is_enabled(),
@@ -234,6 +243,23 @@ pub fn parse_blocklist(text: &str) -> HashSet<String> {
     domains
 }
 
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn heap_bytes_grows_with_domains() {
+        let mut store = BlocklistStore::new();
+        let empty = store.heap_bytes();
+        let domains: HashSet<String> = ["example.com", "example.org", "test.net"]
+            .iter()
+            .map(|s| s.to_string())
+            .collect();
+        store.swap_domains(domains, vec![]);
+        assert!(store.heap_bytes() > empty);
+    }
+}
+
 pub async fn download_blocklists(lists: &[String]) -> Vec<(String, String)> {
     let client = reqwest::Client::builder()
         .timeout(std::time::Duration::from_secs(30))

src/cache.rs

@@ -142,6 +142,26 @@ impl DnsCache {
         self.entry_count = 0;
     }
 
+    pub fn heap_bytes(&self) -> usize {
+        let outer_slot = std::mem::size_of::<u64>()
+            + std::mem::size_of::<String>()
+            + std::mem::size_of::<HashMap<QueryType, CacheEntry>>()
+            + 1;
+        let mut total = self.entries.capacity() * outer_slot;
+        for (domain, type_map) in &self.entries {
+            total += domain.capacity();
+            let inner_slot = std::mem::size_of::<u64>()
+                + std::mem::size_of::<QueryType>()
+                + std::mem::size_of::<CacheEntry>()
+                + 1;
+            total += type_map.capacity() * inner_slot;
+            for entry in type_map.values() {
+                total += entry.packet.heap_bytes();
+            }
+        }
+        total
+    }
+
     pub fn remove(&mut self, domain: &str) {
         let domain_lower = domain.to_lowercase();
         if let Some(type_map) = self.entries.remove(&domain_lower) {
@@ -194,3 +214,23 @@ fn adjust_ttls(records: &mut [DnsRecord], new_ttl: u32) {
         record.set_ttl(new_ttl);
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::packet::DnsPacket;
+
+    #[test]
+    fn heap_bytes_grows_with_entries() {
+        let mut cache = DnsCache::new(100, 1, 3600);
+        let empty = cache.heap_bytes();
+        let mut pkt = DnsPacket::new();
+        pkt.answers.push(DnsRecord::A {
+            domain: "example.com".into(),
+            addr: "1.2.3.4".parse().unwrap(),
+            ttl: 300,
+        });
+        cache.insert("example.com", QueryType::A, &pkt);
+        assert!(cache.heap_bytes() > empty);
+    }
+}

src/config.rs

@@ -1,7 +1,7 @@
 use std::collections::HashMap;
 use std::net::Ipv4Addr;
 use std::net::Ipv6Addr;
-use std::path::Path;
+use std::path::{Path, PathBuf};
 
 use serde::Deserialize;
 
@@ -29,6 +29,8 @@ pub struct Config {
     pub lan: LanConfig,
     #[serde(default)]
     pub dnssec: DnssecConfig,
+    #[serde(default)]
+    pub dot: DotConfig,
 }
 
 #[derive(Deserialize)]
@@ -39,6 +41,10 @@ pub struct ServerConfig {
     pub api_port: u16,
     #[serde(default = "default_api_bind_addr")]
     pub api_bind_addr: String,
+    /// Where numa writes TLS material (CA, leaf certs, regenerated state).
+    /// Defaults to `crate::data_dir()` (platform-specific system path) if unset.
+    #[serde(default)]
+    pub data_dir: Option<PathBuf>,
 }
 
 impl Default for ServerConfig {
@@ -47,6 +53,7 @@ impl Default for ServerConfig {
             bind_addr: default_bind_addr(),
             api_port: default_api_port(),
             api_bind_addr: default_api_bind_addr(),
+            data_dir: None,
         }
     }
 }
@@ -59,18 +66,31 @@ fn default_bind_addr() -> String {
     "0.0.0.0:53".to_string()
 }
 
+pub const DEFAULT_API_PORT: u16 = 5380;
+
 fn default_api_port() -> u16 {
-    5380
+    DEFAULT_API_PORT
 }
 
 #[derive(Deserialize, Default, PartialEq, Eq, Clone, Copy)]
 #[serde(rename_all = "lowercase")]
 pub enum UpstreamMode {
+    Auto,
     #[default]
     Forward,
     Recursive,
 }
 
+impl UpstreamMode {
+    pub fn as_str(&self) -> &'static str {
+        match self {
+            UpstreamMode::Auto => "auto",
+            UpstreamMode::Forward => "forward",
+            UpstreamMode::Recursive => "recursive",
+        }
+    }
+}
+
 #[derive(Deserialize)]
 pub struct UpstreamConfig {
     #[serde(default)]
@@ -103,10 +123,14 @@ impl Default for UpstreamConfig {
     }
 }
 
-fn default_srtt() -> bool {
+fn default_true() -> bool {
     true
 }
 
+fn default_srtt() -> bool {
+    default_true()
+}
+
 fn default_prime_tlds() -> Vec<String> {
     vec![
         // gTLDs
@@ -353,6 +377,41 @@ pub struct DnssecConfig {
     pub strict: bool,
 }
 
+#[derive(Deserialize, Clone)]
+pub struct DotConfig {
+    #[serde(default)]
+    pub enabled: bool,
+    #[serde(default = "default_dot_port")]
+    pub port: u16,
+    #[serde(default = "default_dot_bind_addr")]
+    pub bind_addr: String,
+    /// Path to TLS certificate (PEM). If None, uses self-signed CA.
+    #[serde(default)]
+    pub cert_path: Option<PathBuf>,
+    /// Path to TLS private key (PEM). If None, uses self-signed CA.
+    #[serde(default)]
+    pub key_path: Option<PathBuf>,
+}
+
+impl Default for DotConfig {
+    fn default() -> Self {
+        DotConfig {
+            enabled: false,
+            port: default_dot_port(),
+            bind_addr: default_dot_bind_addr(),
+            cert_path: None,
+            key_path: None,
+        }
+    }
+}
+
+fn default_dot_port() -> u16 {
+    853
+}
+fn default_dot_bind_addr() -> String {
+    "0.0.0.0".to_string()
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;

src/ctx.rs

@@ -62,24 +62,21 @@ pub struct ServerCtx {
     pub dnssec_strict: bool,
 }
 
-pub async fn handle_query(
-    mut buffer: BytePacketBuffer,
+/// Transport-agnostic DNS resolution. Runs the full pipeline (overrides, blocklist,
+/// cache, upstream, DNSSEC) and returns the serialized response in a buffer.
+/// Callers use `.filled()` to get the response bytes without heap allocation.
+/// Callers are responsible for parsing the incoming buffer into a `DnsPacket`
+/// (and logging parse errors) before calling this function.
+pub async fn resolve_query(
+    query: DnsPacket,
     src_addr: SocketAddr,
     ctx: &ServerCtx,
-) -> crate::Result<()> {
+) -> crate::Result<BytePacketBuffer> {
     let start = Instant::now();
 
-    let query = match DnsPacket::from_buffer(&mut buffer) {
-        Ok(packet) => packet,
-        Err(e) => {
-            warn!("{} | PARSE ERROR | {}", src_addr, e);
-            return Ok(());
-        }
-    };
-
     let (qname, qtype) = match query.questions.first() {
         Some(q) => (q.name.clone(), q.qtype),
-        None => return Ok(()),
+        None => return Err("empty question section".into()),
     };
 
     // Pipeline: overrides -> .tld interception -> blocklist -> local zones -> cache -> upstream
@@ -93,18 +90,13 @@ pub async fn handle_query(
         } else if qname == "localhost" || qname.ends_with(".localhost") {
             // RFC 6761: .localhost always resolves to loopback
             let mut resp = DnsPacket::response_from(&query, ResultCode::NOERROR);
-            match qtype {
-                QueryType::AAAA => resp.answers.push(DnsRecord::AAAA {
-                    domain: qname.clone(),
-                    addr: std::net::Ipv6Addr::LOCALHOST,
-                    ttl: 300,
-                }),
-                _ => resp.answers.push(DnsRecord::A {
-                    domain: qname.clone(),
-                    addr: std::net::Ipv4Addr::LOCALHOST,
-                    ttl: 300,
-                }),
-            }
+            resp.answers.push(sinkhole_record(
+                &qname,
+                qtype,
+                std::net::Ipv4Addr::LOCALHOST,
+                std::net::Ipv6Addr::LOCALHOST,
+                300,
+            ));
             (resp, QueryPath::Local, DnssecStatus::Indeterminate)
         } else if is_special_use_domain(&qname) {
             // RFC 6761/8880: private PTR, DDR, NAT64 — answer locally
@@ -113,12 +105,17 @@ pub async fn handle_query(
         } else if !ctx.proxy_tld_suffix.is_empty()
             && (qname.ends_with(&ctx.proxy_tld_suffix) || qname == ctx.proxy_tld)
         {
-            // Resolve .numa: local services → 127.0.0.1, LAN peers → peer IP
+            // Resolve .numa: remote clients get LAN IP (can't reach 127.0.0.1), local get loopback
             let service_name = qname.strip_suffix(&ctx.proxy_tld_suffix).unwrap_or(&qname);
+            let is_remote = !src_addr.ip().is_loopback();
             let resolve_ip = {
                 let local = ctx.services.lock().unwrap();
                 if local.lookup(service_name).is_some() {
-                    std::net::Ipv4Addr::LOCALHOST
+                    if is_remote {
+                        *ctx.lan_ip.lock().unwrap()
+                    } else {
+                        std::net::Ipv4Addr::LOCALHOST
+                    }
                 } else {
                     let mut peers = ctx.lan_peers.lock().unwrap();
                     peers
@@ -130,38 +127,24 @@ pub async fn handle_query(
                         .unwrap_or(std::net::Ipv4Addr::LOCALHOST)
                 }
             };
+            let v6 = if resolve_ip == std::net::Ipv4Addr::LOCALHOST {
+                std::net::Ipv6Addr::LOCALHOST
+            } else {
+                resolve_ip.to_ipv6_mapped()
+            };
             let mut resp = DnsPacket::response_from(&query, ResultCode::NOERROR);
-            match qtype {
-                QueryType::AAAA => resp.answers.push(DnsRecord::AAAA {
-                    domain: qname.clone(),
-                    addr: if resolve_ip == std::net::Ipv4Addr::LOCALHOST {
-                        std::net::Ipv6Addr::LOCALHOST
-                    } else {
-                        resolve_ip.to_ipv6_mapped()
-                    },
-                    ttl: 300,
-                }),
-                _ => resp.answers.push(DnsRecord::A {
-                    domain: qname.clone(),
-                    addr: resolve_ip,
-                    ttl: 300,
-                }),
-            }
+            resp.answers
+                .push(sinkhole_record(&qname, qtype, resolve_ip, v6, 300));
             (resp, QueryPath::Local, DnssecStatus::Indeterminate)
         } else if ctx.blocklist.read().unwrap().is_blocked(&qname) {
             let mut resp = DnsPacket::response_from(&query, ResultCode::NOERROR);
-            match qtype {
-                QueryType::AAAA => resp.answers.push(DnsRecord::AAAA {
-                    domain: qname.clone(),
-                    addr: std::net::Ipv6Addr::UNSPECIFIED,
-                    ttl: 60,
-                }),
-                _ => resp.answers.push(DnsRecord::A {
-                    domain: qname.clone(),
-                    addr: std::net::Ipv4Addr::UNSPECIFIED,
-                    ttl: 60,
-                }),
-            }
+            resp.answers.push(sinkhole_record(
+                &qname,
+                qtype,
+                std::net::Ipv4Addr::UNSPECIFIED,
+                std::net::Ipv6Addr::UNSPECIFIED,
+                60,
+            ));
             (resp, QueryPath::Blocked, DnssecStatus::Indeterminate)
         } else if let Some(records) = ctx.zone_map.get(qname.as_str()).and_then(|m| m.get(&qtype)) {
             let mut resp = DnsPacket::response_from(&query, ResultCode::NOERROR);
@@ -176,64 +159,54 @@ pub async fn handle_query(
                     resp.header.authed_data = true;
                 }
                 (resp, QueryPath::Cached, cached_dnssec)
-            } else if ctx.upstream_mode == UpstreamMode::Recursive {
-                let key = (qname.clone(), qtype);
-                let disposition = acquire_inflight(&ctx.inflight, key.clone());
-
-                match disposition {
-                    Disposition::Follower(mut rx) => {
-                        debug!("{} | {:?} {} | COALESCED", src_addr, qtype, qname);
-                        match rx.recv().await {
-                            Ok(Some(mut resp)) => {
-                                resp.header.id = query.header.id;
-                                (resp, QueryPath::Coalesced, DnssecStatus::Indeterminate)
-                            }
-                            _ => (
-                                DnsPacket::response_from(&query, ResultCode::SERVFAIL),
-                                QueryPath::UpstreamError,
-                                DnssecStatus::Indeterminate,
-                            ),
-                        }
+            } else if let Some(fwd_addr) =
+                crate::system_dns::match_forwarding_rule(&qname, &ctx.forwarding_rules)
+            {
+                // Conditional forwarding takes priority over recursive mode
+                // (e.g. Tailscale .ts.net, VPC private zones)
+                let upstream = Upstream::Udp(fwd_addr);
+                match forward_query(&query, &upstream, ctx.timeout).await {
+                    Ok(resp) => {
+                        ctx.cache.write().unwrap().insert(&qname, qtype, &resp);
+                        (resp, QueryPath::Forwarded, DnssecStatus::Indeterminate)
                     }
-                    Disposition::Leader(tx) => {
-                        // Drop guard: remove inflight entry even on panic/cancellation
-                        let guard = InflightGuard {
-                            inflight: &ctx.inflight,
-                            key: key.clone(),
-                        };
-
-                        let result = crate::recursive::resolve_recursive(
-                            &qname,
-                            qtype,
-                            &ctx.cache,
-                            &query,
-                            &ctx.root_hints,
-                            &ctx.srtt,
+                    Err(e) => {
+                        error!(
+                            "{} | {:?} {} | FORWARD ERROR | {}",
+                            src_addr, qtype, qname, e
+                        );
+                        (
+                            DnsPacket::response_from(&query, ResultCode::SERVFAIL),
+                            QueryPath::UpstreamError,
+                            DnssecStatus::Indeterminate,
                         )
-                        .await;
-
-                        drop(guard);
-
-                        match result {
-                            Ok(resp) => {
-                                let _ = tx.send(Some(resp.clone()));
-                                (resp, QueryPath::Recursive, DnssecStatus::Indeterminate)
-                            }
-                            Err(e) => {
-                                let _ = tx.send(None);
-                                error!(
-                                    "{} | {:?} {} | RECURSIVE ERROR | {}",
-                                    src_addr, qtype, qname, e
-                                );
-                                (
-                                    DnsPacket::response_from(&query, ResultCode::SERVFAIL),
-                                    QueryPath::UpstreamError,
-                                    DnssecStatus::Indeterminate,
-                                )
-                            }
-                        }
                     }
                 }
+            } else if ctx.upstream_mode == UpstreamMode::Recursive {
+                let key = (qname.clone(), qtype);
+                let (resp, path, err) = resolve_coalesced(&ctx.inflight, key, &query, || {
+                    crate::recursive::resolve_recursive(
+                        &qname,
+                        qtype,
+                        &ctx.cache,
+                        &query,
+                        &ctx.root_hints,
+                        &ctx.srtt,
+                    )
+                })
+                .await;
+                if path == QueryPath::Coalesced {
+                    debug!("{} | {:?} {} | COALESCED", src_addr, qtype, qname);
+                } else if path == QueryPath::UpstreamError {
+                    error!(
+                        "{} | {:?} {} | RECURSIVE ERROR | {}",
+                        src_addr,
+                        qtype,
+                        qname,
+                        err.as_deref().unwrap_or("leader failed")
+                    );
+                }
+                (resp, path, DnssecStatus::Indeterminate)
             } else {
                 let upstream =
                     match crate::system_dns::match_forwarding_rule(&qname, &ctx.forwarding_rules) {
@@ -330,17 +303,17 @@ pub async fn handle_query(
         response.resources.len(),
     );
 
+    // Serialize response
+    // TODO: TC bit is UDP-specific; DoT connections could carry up to 65535 bytes.
+    // Once BytePacketBuffer supports larger buffers, skip truncation for TCP/TLS.
     let mut resp_buffer = BytePacketBuffer::new();
     if response.write(&mut resp_buffer).is_err() {
-        // Response too large for UDP — set TC bit and send header + question only
+        // Response too large — set TC bit and send header + question only
         debug!("response too large, setting TC bit for {}", qname);
         let mut tc_response = DnsPacket::response_from(&query, response.header.rescode);
         tc_response.header.truncated_message = true;
-        let mut tc_buffer = BytePacketBuffer::new();
-        tc_response.write(&mut tc_buffer)?;
-        ctx.socket.send_to(tc_buffer.filled(), src_addr).await?;
-    } else {
-        ctx.socket.send_to(resp_buffer.filled(), src_addr).await?;
+        resp_buffer = BytePacketBuffer::new();
+        tc_response.write(&mut resp_buffer)?;
     }
 
     // Record stats and query log
@@ -363,6 +336,30 @@ pub async fn handle_query(
         dnssec,
     });
 
+    Ok(resp_buffer)
+}
+
+/// Handle a DNS query received over UDP. Thin wrapper around resolve_query.
+pub async fn handle_query(
+    mut buffer: BytePacketBuffer,
+    src_addr: SocketAddr,
+    ctx: &ServerCtx,
+) -> crate::Result<()> {
+    let query = match DnsPacket::from_buffer(&mut buffer) {
+        Ok(packet) => packet,
+        Err(e) => {
+            warn!("{} | PARSE ERROR | {}", src_addr, e);
+            return Ok(());
+        }
+    };
+    match resolve_query(query, src_addr, ctx).await {
+        Ok(resp_buffer) => {
+            ctx.socket.send_to(resp_buffer.filled(), src_addr).await?;
+        }
+        Err(e) => {
+            warn!("{} | RESOLVE ERROR | {}", src_addr, e);
+        }
+    }
     Ok(())
 }
 
@@ -416,6 +413,27 @@ fn is_special_use_domain(qname: &str) -> bool {
     qname == "local" || qname.ends_with(".local")
 }
 
+fn sinkhole_record(
+    domain: &str,
+    qtype: QueryType,
+    v4: std::net::Ipv4Addr,
+    v6: std::net::Ipv6Addr,
+    ttl: u32,
+) -> DnsRecord {
+    match qtype {
+        QueryType::AAAA => DnsRecord::AAAA {
+            domain: domain.to_string(),
+            addr: v6,
+            ttl,
+        },
+        _ => DnsRecord::A {
+            domain: domain.to_string(),
+            addr: v4,
+            ttl,
+        },
+    }
+}
+
 enum Disposition {
     Leader(broadcast::Sender<Option<DnsPacket>>),
     Follower(broadcast::Receiver<Option<DnsPacket>>),
@@ -432,6 +450,57 @@ fn acquire_inflight(inflight: &Mutex<InflightMap>, key: (String, QueryType)) ->
     }
 }
 
+/// Run a resolve function with in-flight coalescing. Multiple concurrent calls
+/// for the same key share a single resolution — the first caller (leader)
+/// executes `resolve_fn`, and followers wait for the broadcast result.
+async fn resolve_coalesced<F, Fut>(
+    inflight: &Mutex<InflightMap>,
+    key: (String, QueryType),
+    query: &DnsPacket,
+    resolve_fn: F,
+) -> (DnsPacket, QueryPath, Option<String>)
+where
+    F: FnOnce() -> Fut,
+    Fut: std::future::Future<Output = crate::Result<DnsPacket>>,
+{
+    let disposition = acquire_inflight(inflight, key.clone());
+
+    match disposition {
+        Disposition::Follower(mut rx) => match rx.recv().await {
+            Ok(Some(mut resp)) => {
+                resp.header.id = query.header.id;
+                (resp, QueryPath::Coalesced, None)
+            }
+            _ => (
+                DnsPacket::response_from(query, ResultCode::SERVFAIL),
+                QueryPath::UpstreamError,
+                None,
+            ),
+        },
+        Disposition::Leader(tx) => {
+            let guard = InflightGuard { inflight, key };
+            let result = resolve_fn().await;
+            drop(guard);
+
+            match result {
+                Ok(resp) => {
+                    let _ = tx.send(Some(resp.clone()));
+                    (resp, QueryPath::Recursive, None)
+                }
+                Err(e) => {
+                    let _ = tx.send(None);
+                    let err_msg = e.to_string();
+                    (
+                        DnsPacket::response_from(query, ResultCode::SERVFAIL),
+                        QueryPath::UpstreamError,
+                        Some(err_msg),
+                    )
+                }
+            }
+        }
+    }
+}
+
 struct InflightGuard<'a> {
     inflight: &'a Mutex<InflightMap>,
     key: (String, QueryType),
@@ -443,20 +512,6 @@ impl Drop for InflightGuard<'_> {
     }
 }
 
-/// Build a wire-format DNS query packet for the given domain and type.
-#[cfg(test)]
-fn build_wire_query(id: u16, domain: &str, qtype: QueryType) -> BytePacketBuffer {
-    let mut pkt = DnsPacket::new();
-    pkt.header.id = id;
-    pkt.header.recursion_desired = true;
-    pkt.header.questions = 1;
-    pkt.questions
-        .push(crate::question::DnsQuestion::new(domain.to_string(), qtype));
-    let mut buf = BytePacketBuffer::new();
-    pkt.write(&mut buf).unwrap();
-    BytePacketBuffer::from_bytes(buf.filled())
-}
-
 fn special_use_response(query: &DnsPacket, qname: &str, qtype: QueryType) -> DnsPacket {
     use std::net::{Ipv4Addr, Ipv6Addr};
     if qname == "ipv4only.arpa" {
@@ -495,8 +550,8 @@ fn special_use_response(query: &DnsPacket, qname: &str, qtype: QueryType) -> Dns
 mod tests {
     use super::*;
     use std::collections::HashMap;
-    use std::net::{Ipv4Addr, SocketAddr};
-    use std::sync::{Arc, Mutex, RwLock};
+    use std::net::Ipv4Addr;
+    use std::sync::{Arc, Mutex};
     use tokio::sync::broadcast;
 
     // ---- InflightGuard unit tests ----
@@ -669,189 +724,212 @@ mod tests {
         }
     }
 
-    // ---- Integration: concurrent handle_query coalescing ----
+    // ---- Integration: resolve_coalesced with mock futures ----
 
-    use tokio::io::{AsyncReadExt, AsyncWriteExt};
-    use tokio::net::TcpListener;
-
-    /// Spawn a slow TCP DNS server that delays `delay` before responding.
-    /// Returns (addr, query_count) where query_count is an Arc<AtomicU32>
-    /// tracking how many queries were actually resolved (not coalesced).
-    async fn spawn_slow_dns_server(
-        delay: Duration,
-    ) -> (SocketAddr, Arc<std::sync::atomic::AtomicU32>) {
-        let listener = TcpListener::bind("127.0.0.1:0").await.unwrap();
-        let addr = listener.local_addr().unwrap();
-        let count = Arc::new(std::sync::atomic::AtomicU32::new(0));
-        let count_clone = count.clone();
-
-        tokio::spawn(async move {
-            loop {
-                let (mut stream, _) = match listener.accept().await {
-                    Ok(c) => c,
-                    Err(_) => break,
-                };
-                let count = count_clone.clone();
-                let delay = delay;
-                tokio::spawn(async move {
-                    let mut len_buf = [0u8; 2];
-                    if stream.read_exact(&mut len_buf).await.is_err() {
-                        return;
-                    }
-                    let len = u16::from_be_bytes(len_buf) as usize;
-                    let mut data = vec![0u8; len];
-                    if stream.read_exact(&mut data).await.is_err() {
-                        return;
-                    }
-
-                    let mut buf = BytePacketBuffer::from_bytes(&data);
-                    let query = match DnsPacket::from_buffer(&mut buf) {
-                        Ok(q) => q,
-                        Err(_) => return,
-                    };
-
-                    count.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
-
-                    // Deliberate delay to create coalescing window
-                    tokio::time::sleep(delay).await;
-
-                    let mut resp = DnsPacket::response_from(&query, ResultCode::NOERROR);
-                    resp.header.authoritative_answer = true;
-                    if let Some(q) = query.questions.first() {
-                        resp.answers.push(DnsRecord::A {
-                            domain: q.name.clone(),
-                            addr: Ipv4Addr::new(10, 0, 0, 1),
-                            ttl: 300,
-                        });
-                    }
-
-                    let mut resp_buf = BytePacketBuffer::new();
-                    if resp.write(&mut resp_buf).is_err() {
-                        return;
-                    }
-                    let resp_bytes = resp_buf.filled();
-                    let mut out = Vec::with_capacity(2 + resp_bytes.len());
-                    out.extend_from_slice(&(resp_bytes.len() as u16).to_be_bytes());
-                    out.extend_from_slice(resp_bytes);
-                    let _ = stream.write_all(&out).await;
-                });
-            }
+    fn mock_response(domain: &str) -> DnsPacket {
+        let mut resp = DnsPacket::new();
+        resp.header.response = true;
+        resp.header.rescode = ResultCode::NOERROR;
+        resp.answers.push(DnsRecord::A {
+            domain: domain.to_string(),
+            addr: Ipv4Addr::new(10, 0, 0, 1),
+            ttl: 300,
         });
-        (addr, count)
-    }
-
-    async fn test_recursive_ctx(root_hint: SocketAddr) -> Arc<ServerCtx> {
-        let socket = tokio::net::UdpSocket::bind("127.0.0.1:0").await.unwrap();
-        Arc::new(ServerCtx {
-            socket,
-            zone_map: HashMap::new(),
-            cache: RwLock::new(crate::cache::DnsCache::new(100, 60, 86400)),
-            stats: Mutex::new(crate::stats::ServerStats::new()),
-            overrides: RwLock::new(crate::override_store::OverrideStore::new()),
-            blocklist: RwLock::new(crate::blocklist::BlocklistStore::new()),
-            query_log: Mutex::new(crate::query_log::QueryLog::new(100)),
-            services: Mutex::new(crate::service_store::ServiceStore::new()),
-            lan_peers: Mutex::new(crate::lan::PeerStore::new(90)),
-            forwarding_rules: Vec::new(),
-            upstream: Mutex::new(crate::forward::Upstream::Udp(
-                "127.0.0.1:53".parse().unwrap(),
-            )),
-            upstream_auto: false,
-            upstream_port: 53,
-            lan_ip: Mutex::new(Ipv4Addr::LOCALHOST),
-            timeout: Duration::from_secs(3),
-            proxy_tld: "numa".to_string(),
-            proxy_tld_suffix: ".numa".to_string(),
-            lan_enabled: false,
-            config_path: "/tmp/test-numa.toml".to_string(),
-            config_found: false,
-            config_dir: std::path::PathBuf::from("/tmp"),
-            data_dir: std::path::PathBuf::from("/tmp"),
-            tls_config: None,
-            upstream_mode: crate::config::UpstreamMode::Recursive,
-            root_hints: vec![root_hint],
-            srtt: RwLock::new(crate::srtt::SrttCache::new(true)),
-            inflight: Mutex::new(HashMap::new()),
-            dnssec_enabled: false,
-            dnssec_strict: false,
-        })
+        resp
     }
 
     #[tokio::test]
     async fn concurrent_queries_coalesce_to_single_resolution() {
-        // Force TCP-only so mock server works
-        crate::recursive::UDP_DISABLED.store(true, std::sync::atomic::Ordering::Release);
+        let inflight = Arc::new(Mutex::new(HashMap::new()));
+        let resolve_count = Arc::new(std::sync::atomic::AtomicU32::new(0));
 
-        let (server_addr, query_count) = spawn_slow_dns_server(Duration::from_millis(200)).await;
-        let ctx = test_recursive_ctx(server_addr).await;
-        let src: SocketAddr = "127.0.0.1:9999".parse().unwrap();
-
-        // Fire 5 concurrent queries for the same (domain, A)
         let mut handles = Vec::new();
         for i in 0..5u16 {
-            let ctx = ctx.clone();
-            let buf = build_wire_query(100 + i, "coalesce-test.example.com", QueryType::A);
-            handles.push(tokio::spawn(
-                async move { handle_query(buf, src, &ctx).await },
-            ));
+            let count = resolve_count.clone();
+            let inf = inflight.clone();
+            let key = ("coalesce.test".to_string(), QueryType::A);
+            let query = DnsPacket::query(100 + i, "coalesce.test", QueryType::A);
+            handles.push(tokio::spawn(async move {
+                resolve_coalesced(&inf, key, &query, || async {
+                    count.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
+                    tokio::time::sleep(Duration::from_millis(200)).await;
+                    Ok(mock_response("coalesce.test"))
+                })
+                .await
+            }));
         }
 
+        let mut paths = Vec::new();
         for h in handles {
-            h.await.unwrap().unwrap();
+            let (_, path, _) = h.await.unwrap();
+            paths.push(path);
         }
 
-        // Only 1 resolution should have reached the upstream server
-        let actual = query_count.load(std::sync::atomic::Ordering::Relaxed);
-        assert_eq!(actual, 1, "expected 1 upstream query, got {}", actual);
+        let actual = resolve_count.load(std::sync::atomic::Ordering::Relaxed);
+        assert_eq!(actual, 1, "expected 1 resolution, got {}", actual);
 
-        // Inflight map must be empty after all queries complete
-        assert!(ctx.inflight.lock().unwrap().is_empty());
+        let recursive = paths.iter().filter(|p| **p == QueryPath::Recursive).count();
+        let coalesced = paths.iter().filter(|p| **p == QueryPath::Coalesced).count();
+        assert_eq!(recursive, 1, "expected 1 RECURSIVE, got {}", recursive);
+        assert_eq!(coalesced, 4, "expected 4 COALESCED, got {}", coalesced);
 
-        crate::recursive::reset_udp_state();
+        assert!(inflight.lock().unwrap().is_empty());
     }
 
     #[tokio::test]
     async fn different_qtypes_not_coalesced() {
-        crate::recursive::UDP_DISABLED.store(true, std::sync::atomic::Ordering::Release);
+        let inflight = Arc::new(Mutex::new(HashMap::new()));
+        let resolve_count = Arc::new(std::sync::atomic::AtomicU32::new(0));
 
-        let (server_addr, query_count) = spawn_slow_dns_server(Duration::from_millis(100)).await;
-        let ctx = test_recursive_ctx(server_addr).await;
-        let src: SocketAddr = "127.0.0.1:9999".parse().unwrap();
+        let inf1 = inflight.clone();
+        let inf2 = inflight.clone();
+        let count1 = resolve_count.clone();
+        let count2 = resolve_count.clone();
 
-        // Fire A and AAAA concurrently — should NOT coalesce
-        let ctx_ref = ctx.clone();
-        let ctx_ref2 = ctx.clone();
-        let buf_a = build_wire_query(200, "different-qt.example.com", QueryType::A);
-        let buf_aaaa = build_wire_query(201, "different-qt.example.com", QueryType::AAAA);
+        let query_a = DnsPacket::query(200, "same.domain", QueryType::A);
+        let query_aaaa = DnsPacket::query(201, "same.domain", QueryType::AAAA);
 
-        let h1 = tokio::spawn(async move { handle_query(buf_a, src, &ctx_ref).await });
-        let h2 = tokio::spawn(async move { handle_query(buf_aaaa, src, &ctx_ref2).await });
+        let h1 = tokio::spawn(async move {
+            resolve_coalesced(
+                &inf1,
+                ("same.domain".to_string(), QueryType::A),
+                &query_a,
+                || async {
+                    count1.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
+                    tokio::time::sleep(Duration::from_millis(100)).await;
+                    Ok(mock_response("same.domain"))
+                },
+            )
+            .await
+        });
+        let h2 = tokio::spawn(async move {
+            resolve_coalesced(
+                &inf2,
+                ("same.domain".to_string(), QueryType::AAAA),
+                &query_aaaa,
+                || async {
+                    count2.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
+                    tokio::time::sleep(Duration::from_millis(100)).await;
+                    Ok(mock_response("same.domain"))
+                },
+            )
+            .await
+        });
 
-        h1.await.unwrap().unwrap();
-        h2.await.unwrap().unwrap();
+        let (_, path1, _) = h1.await.unwrap();
+        let (_, path2, _) = h2.await.unwrap();
 
-        let actual = query_count.load(std::sync::atomic::Ordering::Relaxed);
-        assert!(
-            actual >= 2,
-            "A and AAAA should resolve independently, got {}",
-            actual
-        );
-        assert!(ctx.inflight.lock().unwrap().is_empty());
+        let actual = resolve_count.load(std::sync::atomic::Ordering::Relaxed);
+        assert_eq!(actual, 2, "A and AAAA should each resolve, got {}", actual);
+        assert_eq!(path1, QueryPath::Recursive);
+        assert_eq!(path2, QueryPath::Recursive);
 
-        crate::recursive::reset_udp_state();
+        assert!(inflight.lock().unwrap().is_empty());
     }
 
     #[tokio::test]
-    async fn inflight_map_cleaned_after_upstream_error() {
-        // Server that rejects everything — no server running at all
-        let bogus_addr: SocketAddr = "127.0.0.1:1".parse().unwrap();
-        let ctx = test_recursive_ctx(bogus_addr).await;
-        let src: SocketAddr = "127.0.0.1:9999".parse().unwrap();
+    async fn inflight_map_cleaned_after_error() {
+        let inflight: Mutex<InflightMap> = Mutex::new(HashMap::new());
+        let query = DnsPacket::query(300, "will-fail.test", QueryType::A);
 
-        let buf = build_wire_query(300, "will-fail.example.com", QueryType::A);
-        let _ = handle_query(buf, src, &ctx).await;
+        let (_, path, _) = resolve_coalesced(
+            &inflight,
+            ("will-fail.test".to_string(), QueryType::A),
+            &query,
+            || async { Err::<DnsPacket, _>("upstream timeout".into()) },
+        )
+        .await;
 
-        // Map must be clean even after error
-        assert!(ctx.inflight.lock().unwrap().is_empty());
+        assert_eq!(path, QueryPath::UpstreamError);
+        assert!(inflight.lock().unwrap().is_empty());
+    }
+
+    #[tokio::test]
+    async fn follower_gets_servfail_when_leader_fails() {
+        let inflight = Arc::new(Mutex::new(HashMap::new()));
+
+        let mut handles = Vec::new();
+        for i in 0..3u16 {
+            let inf = inflight.clone();
+            let query = DnsPacket::query(400 + i, "fail.test", QueryType::A);
+            handles.push(tokio::spawn(async move {
+                resolve_coalesced(
+                    &inf,
+                    ("fail.test".to_string(), QueryType::A),
+                    &query,
+                    || async {
+                        tokio::time::sleep(Duration::from_millis(200)).await;
+                        Err::<DnsPacket, _>("upstream error".into())
+                    },
+                )
+                .await
+            }));
+        }
+
+        let mut paths = Vec::new();
+        for h in handles {
+            let (resp, path, _) = h.await.unwrap();
+            assert_eq!(resp.header.rescode, ResultCode::SERVFAIL);
+            assert_eq!(
+                resp.questions.len(),
+                1,
+                "SERVFAIL must echo question section"
+            );
+            assert_eq!(resp.questions[0].name, "fail.test");
+            paths.push(path);
+        }
+
+        let errors = paths
+            .iter()
+            .filter(|p| **p == QueryPath::UpstreamError)
+            .count();
+        assert_eq!(errors, 3, "all 3 should be UpstreamError, got {}", errors);
+
+        assert!(inflight.lock().unwrap().is_empty());
+    }
+
+    #[tokio::test]
+    async fn servfail_leader_includes_question_section() {
+        let inflight: Mutex<InflightMap> = Mutex::new(HashMap::new());
+        let query = DnsPacket::query(500, "question.test", QueryType::A);
+
+        let (resp, _, _) = resolve_coalesced(
+            &inflight,
+            ("question.test".to_string(), QueryType::A),
+            &query,
+            || async { Err::<DnsPacket, _>("fail".into()) },
+        )
+        .await;
+
+        assert_eq!(resp.header.rescode, ResultCode::SERVFAIL);
+        assert_eq!(
+            resp.questions.len(),
+            1,
+            "SERVFAIL must echo question section"
+        );
+        assert_eq!(resp.questions[0].name, "question.test");
+        assert_eq!(resp.questions[0].qtype, QueryType::A);
+        assert_eq!(resp.header.id, 500);
+    }
+
+    #[tokio::test]
+    async fn leader_error_preserves_message() {
+        let inflight: Mutex<InflightMap> = Mutex::new(HashMap::new());
+        let query = DnsPacket::query(700, "err-msg.test", QueryType::A);
+
+        let (_, path, err) = resolve_coalesced(
+            &inflight,
+            ("err-msg.test".to_string(), QueryType::A),
+            &query,
+            || async { Err::<DnsPacket, _>("connection refused by upstream".into()) },
+        )
+        .await;
+
+        assert_eq!(path, QueryPath::UpstreamError);
+        assert_eq!(
+            err.as_deref(),
+            Some("connection refused by upstream"),
+            "error message must be preserved for logging"
+        );
     }
 }

src/dot.rs

@@ -0,0 +1,542 @@
+use std::net::{IpAddr, SocketAddr};
+use std::path::Path;
+use std::sync::Arc;
+use std::time::Duration;
+
+use log::{debug, error, info, warn};
+use rustls::ServerConfig;
+use tokio::io::{AsyncReadExt, AsyncWriteExt};
+use tokio::net::TcpListener;
+use tokio::sync::Semaphore;
+use tokio_rustls::TlsAcceptor;
+
+use crate::buffer::BytePacketBuffer;
+use crate::config::DotConfig;
+use crate::ctx::{resolve_query, ServerCtx};
+use crate::header::ResultCode;
+use crate::packet::DnsPacket;
+
+const MAX_CONNECTIONS: usize = 512;
+const IDLE_TIMEOUT: Duration = Duration::from_secs(30);
+const HANDSHAKE_TIMEOUT: Duration = Duration::from_secs(10);
+const WRITE_TIMEOUT: Duration = Duration::from_secs(10);
+// Matches BytePacketBuffer::BUF_SIZE — RFC 7858 allows up to 65535 but our
+// buffer would silently truncate anything larger.
+const MAX_MSG_LEN: usize = 4096;
+
+fn dot_alpn() -> Vec<Vec<u8>> {
+    vec![b"dot".to_vec()]
+}
+
+/// Build a TLS ServerConfig for DoT from user-provided cert/key PEM files.
+fn load_tls_config(cert_path: &Path, key_path: &Path) -> crate::Result<Arc<ServerConfig>> {
+    // rustls needs a CryptoProvider installed before ServerConfig::builder().
+    // The proxy's build_tls_config also does this; we repeat it here because
+    // running DoT with user-provided certs while the proxy is disabled would
+    // otherwise panic on first handshake (no default provider).
+    let _ = rustls::crypto::ring::default_provider().install_default();
+
+    let cert_pem = std::fs::read(cert_path)?;
+    let key_pem = std::fs::read(key_path)?;
+
+    let certs: Vec<_> = rustls_pemfile::certs(&mut &cert_pem[..]).collect::<Result<_, _>>()?;
+    let key = rustls_pemfile::private_key(&mut &key_pem[..])?
+        .ok_or("no private key found in key file")?;
+
+    let mut config = ServerConfig::builder()
+        .with_no_client_auth()
+        .with_single_cert(certs, key)?;
+    config.alpn_protocols = dot_alpn();
+
+    Ok(Arc::new(config))
+}
+
+/// Build a self-signed DoT TLS config. Can't reuse `ctx.tls_config` (the
+/// proxy's shared config) because DoT needs its own ALPN advertisement.
+///
+/// Pass `proxy_tld` itself as a service name so the cert gets an explicit
+/// `{tld}.{tld}` SAN (e.g. "numa.numa") matching the ServerName that
+/// setup-phone's mobileconfig sends as SNI. The `*.{tld}` wildcard alone
+/// is rejected by strict TLS clients under single-label TLDs (per the
+/// note in tls.rs::generate_service_cert).
+fn self_signed_tls(ctx: &ServerCtx) -> Option<Arc<ServerConfig>> {
+    let service_names = [ctx.proxy_tld.clone()];
+    match crate::tls::build_tls_config(&ctx.proxy_tld, &service_names, dot_alpn(), &ctx.data_dir) {
+        Ok(cfg) => Some(cfg),
+        Err(e) => {
+            warn!(
+                "DoT: failed to generate self-signed TLS: {} — DoT disabled",
+                e
+            );
+            None
+        }
+    }
+}
+
+/// Start the DNS-over-TLS listener (RFC 7858).
+pub async fn start_dot(ctx: Arc<ServerCtx>, config: &DotConfig) {
+    let tls_config = match (&config.cert_path, &config.key_path) {
+        (Some(cert), Some(key)) => match load_tls_config(cert, key) {
+            Ok(cfg) => cfg,
+            Err(e) => {
+                warn!("DoT: failed to load TLS cert/key: {} — DoT disabled", e);
+                return;
+            }
+        },
+        _ => match self_signed_tls(&ctx) {
+            Some(cfg) => cfg,
+            None => return,
+        },
+    };
+
+    let bind_addr: IpAddr = config
+        .bind_addr
+        .parse()
+        .unwrap_or(IpAddr::V4(std::net::Ipv4Addr::UNSPECIFIED));
+    let addr = SocketAddr::new(bind_addr, config.port);
+    let listener = match TcpListener::bind(addr).await {
+        Ok(l) => l,
+        Err(e) => {
+            warn!("DoT: could not bind {} ({}) — DoT disabled", addr, e);
+            return;
+        }
+    };
+    info!("DoT listening on {}", addr);
+
+    accept_loop(listener, TlsAcceptor::from(tls_config), ctx).await;
+}
+
+async fn accept_loop(listener: TcpListener, acceptor: TlsAcceptor, ctx: Arc<ServerCtx>) {
+    let semaphore = Arc::new(Semaphore::new(MAX_CONNECTIONS));
+
+    loop {
+        let (tcp_stream, remote_addr) = match listener.accept().await {
+            Ok(conn) => conn,
+            Err(e) => {
+                error!("DoT: TCP accept error: {}", e);
+                // Back off to avoid tight-looping on persistent failures (e.g. fd exhaustion).
+                tokio::time::sleep(Duration::from_millis(100)).await;
+                continue;
+            }
+        };
+
+        let permit = match semaphore.clone().try_acquire_owned() {
+            Ok(p) => p,
+            Err(_) => {
+                debug!("DoT: connection limit reached, rejecting {}", remote_addr);
+                continue;
+            }
+        };
+        let acceptor = acceptor.clone();
+        let ctx = Arc::clone(&ctx);
+
+        tokio::spawn(async move {
+            let _permit = permit; // held until task exits
+
+            let tls_stream =
+                match tokio::time::timeout(HANDSHAKE_TIMEOUT, acceptor.accept(tcp_stream)).await {
+                    Ok(Ok(s)) => s,
+                    Ok(Err(e)) => {
+                        debug!("DoT: TLS handshake failed from {}: {}", remote_addr, e);
+                        return;
+                    }
+                    Err(_) => {
+                        debug!("DoT: TLS handshake timeout from {}", remote_addr);
+                        return;
+                    }
+                };
+
+            handle_dot_connection(tls_stream, remote_addr, &ctx).await;
+        });
+    }
+}
+
+/// Handle a single persistent DoT connection (RFC 7858).
+/// Reads length-prefixed DNS queries until EOF, idle timeout, or error.
+async fn handle_dot_connection<S>(mut stream: S, remote_addr: SocketAddr, ctx: &ServerCtx)
+where
+    S: AsyncReadExt + AsyncWriteExt + Unpin,
+{
+    loop {
+        // Read 2-byte length prefix (RFC 1035 §4.2.2) with idle timeout
+        let mut len_buf = [0u8; 2];
+        let Ok(Ok(_)) = tokio::time::timeout(IDLE_TIMEOUT, stream.read_exact(&mut len_buf)).await
+        else {
+            break;
+        };
+        let msg_len = u16::from_be_bytes(len_buf) as usize;
+        if msg_len > MAX_MSG_LEN {
+            debug!("DoT: oversized message {} from {}", msg_len, remote_addr);
+            break;
+        }
+
+        let mut buffer = BytePacketBuffer::new();
+        let Ok(Ok(_)) =
+            tokio::time::timeout(IDLE_TIMEOUT, stream.read_exact(&mut buffer.buf[..msg_len])).await
+        else {
+            break;
+        };
+
+        // Parse query up-front so we can echo its question section in SERVFAIL
+        // responses when resolve_query fails.
+        let query = match DnsPacket::from_buffer(&mut buffer) {
+            Ok(q) => q,
+            Err(e) => {
+                warn!("{} | PARSE ERROR | {}", remote_addr, e);
+                // BytePacketBuffer is zero-initialized, so buf[0..2] reads as 0x0000
+                // for sub-2-byte messages — harmless FORMERR with id=0.
+                let query_id = u16::from_be_bytes([buffer.buf[0], buffer.buf[1]]);
+                let mut resp = DnsPacket::new();
+                resp.header.id = query_id;
+                resp.header.response = true;
+                resp.header.rescode = ResultCode::FORMERR;
+                if send_response(&mut stream, &resp, remote_addr)
+                    .await
+                    .is_err()
+                {
+                    break;
+                }
+                continue;
+            }
+        };
+
+        match resolve_query(query.clone(), remote_addr, ctx).await {
+            Ok(resp_buffer) => {
+                if write_framed(&mut stream, resp_buffer.filled())
+                    .await
+                    .is_err()
+                {
+                    break;
+                }
+            }
+            Err(e) => {
+                warn!("{} | RESOLVE ERROR | {}", remote_addr, e);
+                // SERVFAIL that echoes the original question section.
+                let resp = DnsPacket::response_from(&query, ResultCode::SERVFAIL);
+                if send_response(&mut stream, &resp, remote_addr)
+                    .await
+                    .is_err()
+                {
+                    break;
+                }
+            }
+        }
+    }
+}
+
+/// Serialize a DNS response and send it framed. Logs serialization failures
+/// and returns Err so the caller can tear down the connection.
+async fn send_response<S>(
+    stream: &mut S,
+    resp: &DnsPacket,
+    remote_addr: SocketAddr,
+) -> std::io::Result<()>
+where
+    S: AsyncWriteExt + Unpin,
+{
+    let mut out_buf = BytePacketBuffer::new();
+    if resp.write(&mut out_buf).is_err() {
+        debug!(
+            "DoT: failed to serialize {:?} response for {}",
+            resp.header.rescode, remote_addr
+        );
+        return Err(std::io::Error::other("serialize failed"));
+    }
+    write_framed(stream, out_buf.filled()).await
+}
+
+/// Write a DNS message with its 2-byte length prefix, coalesced into one syscall.
+/// Bounded by WRITE_TIMEOUT so a stalled reader can't indefinitely hold a worker.
+async fn write_framed<S>(stream: &mut S, msg: &[u8]) -> std::io::Result<()>
+where
+    S: AsyncWriteExt + Unpin,
+{
+    let mut out = Vec::with_capacity(2 + msg.len());
+    out.extend_from_slice(&(msg.len() as u16).to_be_bytes());
+    out.extend_from_slice(msg);
+    match tokio::time::timeout(WRITE_TIMEOUT, async {
+        stream.write_all(&out).await?;
+        stream.flush().await
+    })
+    .await
+    {
+        Ok(result) => result,
+        Err(_) => Err(std::io::Error::other("write timeout")),
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use std::collections::HashMap;
+    use std::sync::{Mutex, RwLock};
+
+    use rcgen::{CertificateParams, DnType, KeyPair};
+    use rustls::pki_types::{CertificateDer, PrivateKeyDer, PrivatePkcs8KeyDer, ServerName};
+    use tokio::io::{AsyncReadExt, AsyncWriteExt};
+
+    use crate::buffer::BytePacketBuffer;
+    use crate::header::ResultCode;
+    use crate::packet::DnsPacket;
+    use crate::question::QueryType;
+    use crate::record::DnsRecord;
+
+    /// Generate a self-signed DoT server config and return its leaf cert DER
+    /// so callers can build matching client configs with arbitrary ALPN.
+    fn test_tls_configs() -> (Arc<ServerConfig>, CertificateDer<'static>) {
+        let _ = rustls::crypto::ring::default_provider().install_default();
+
+        // Mirror production self_signed_tls SAN shape: *.numa wildcard plus
+        // explicit numa.numa apex (the ServerName setup-phone uses as SNI).
+        let key_pair = KeyPair::generate().unwrap();
+        let mut params = CertificateParams::default();
+        params
+            .distinguished_name
+            .push(DnType::CommonName, "Numa .numa services");
+        params.subject_alt_names = vec![
+            rcgen::SanType::DnsName("*.numa".try_into().unwrap()),
+            rcgen::SanType::DnsName("numa.numa".try_into().unwrap()),
+        ];
+        let cert = params.self_signed(&key_pair).unwrap();
+
+        let cert_der = CertificateDer::from(cert.der().to_vec());
+        let key_der = PrivateKeyDer::Pkcs8(PrivatePkcs8KeyDer::from(key_pair.serialize_der()));
+
+        let mut server_config = ServerConfig::builder()
+            .with_no_client_auth()
+            .with_single_cert(vec![cert_der.clone()], key_der)
+            .unwrap();
+        server_config.alpn_protocols = dot_alpn();
+
+        (Arc::new(server_config), cert_der)
+    }
+
+    /// Build a TLS client config that trusts `cert_der` and advertises the
+    /// given ALPN protocols. Used by tests to vary ALPN per test case.
+    fn dot_client(
+        cert_der: &CertificateDer<'static>,
+        alpn: Vec<Vec<u8>>,
+    ) -> Arc<rustls::ClientConfig> {
+        let mut root_store = rustls::RootCertStore::empty();
+        root_store.add(cert_der.clone()).unwrap();
+        let mut config = rustls::ClientConfig::builder()
+            .with_root_certificates(root_store)
+            .with_no_client_auth();
+        config.alpn_protocols = alpn;
+        Arc::new(config)
+    }
+
+    /// Spin up a DoT listener with a test TLS config. Returns the bind addr
+    /// and the leaf cert DER so callers can build clients with arbitrary ALPN.
+    /// The upstream is pointed at a bound-but-unresponsive UDP socket we own, so
+    /// any query that escapes to the upstream path times out deterministically
+    /// (SERVFAIL) regardless of what the host has running on port 53.
+    async fn spawn_dot_server() -> (SocketAddr, CertificateDer<'static>) {
+        let (server_tls, cert_der) = test_tls_configs();
+
+        let socket = tokio::net::UdpSocket::bind("127.0.0.1:0").await.unwrap();
+        // Bind an unresponsive upstream and leak it so it lives for the test duration.
+        let blackhole = Box::leak(Box::new(std::net::UdpSocket::bind("127.0.0.1:0").unwrap()));
+        let upstream_addr = blackhole.local_addr().unwrap();
+        let ctx = Arc::new(ServerCtx {
+            socket,
+            zone_map: {
+                let mut m = HashMap::new();
+                let mut inner = HashMap::new();
+                inner.insert(
+                    QueryType::A,
+                    vec![DnsRecord::A {
+                        domain: "dot-test.example".to_string(),
+                        addr: std::net::Ipv4Addr::new(10, 0, 0, 1),
+                        ttl: 300,
+                    }],
+                );
+                m.insert("dot-test.example".to_string(), inner);
+                m
+            },
+            cache: RwLock::new(crate::cache::DnsCache::new(100, 60, 86400)),
+            stats: Mutex::new(crate::stats::ServerStats::new()),
+            overrides: RwLock::new(crate::override_store::OverrideStore::new()),
+            blocklist: RwLock::new(crate::blocklist::BlocklistStore::new()),
+            query_log: Mutex::new(crate::query_log::QueryLog::new(100)),
+            services: Mutex::new(crate::service_store::ServiceStore::new()),
+            lan_peers: Mutex::new(crate::lan::PeerStore::new(90)),
+            forwarding_rules: Vec::new(),
+            upstream: Mutex::new(crate::forward::Upstream::Udp(upstream_addr)),
+            upstream_auto: false,
+            upstream_port: 53,
+            lan_ip: Mutex::new(std::net::Ipv4Addr::LOCALHOST),
+            timeout: Duration::from_millis(200),
+            proxy_tld: "numa".to_string(),
+            proxy_tld_suffix: ".numa".to_string(),
+            lan_enabled: false,
+            config_path: String::new(),
+            config_found: false,
+            config_dir: std::path::PathBuf::from("/tmp"),
+            data_dir: std::path::PathBuf::from("/tmp"),
+            tls_config: Some(arc_swap::ArcSwap::from(server_tls)),
+            upstream_mode: crate::config::UpstreamMode::Forward,
+            root_hints: Vec::new(),
+            srtt: RwLock::new(crate::srtt::SrttCache::new(true)),
+            inflight: Mutex::new(HashMap::new()),
+            dnssec_enabled: false,
+            dnssec_strict: false,
+        });
+
+        let listener = TcpListener::bind("127.0.0.1:0").await.unwrap();
+        let addr = listener.local_addr().unwrap();
+
+        let tls_config = Arc::clone(&*ctx.tls_config.as_ref().unwrap().load());
+        let acceptor = TlsAcceptor::from(tls_config);
+
+        tokio::spawn(accept_loop(listener, acceptor, ctx));
+
+        (addr, cert_der)
+    }
+
+    /// Open a TLS connection to the DoT server and return the stream.
+    /// Uses SNI "numa.numa" to mirror what setup-phone's mobileconfig sends.
+    async fn dot_connect(
+        addr: SocketAddr,
+        client_config: &Arc<rustls::ClientConfig>,
+    ) -> tokio_rustls::client::TlsStream<tokio::net::TcpStream> {
+        let connector = tokio_rustls::TlsConnector::from(Arc::clone(client_config));
+        let tcp = tokio::net::TcpStream::connect(addr).await.unwrap();
+        connector
+            .connect(ServerName::try_from("numa.numa").unwrap(), tcp)
+            .await
+            .unwrap()
+    }
+
+    /// Send a DNS query over a DoT stream and read the response.
+    async fn dot_exchange(
+        stream: &mut tokio_rustls::client::TlsStream<tokio::net::TcpStream>,
+        query: &DnsPacket,
+    ) -> DnsPacket {
+        let mut buf = BytePacketBuffer::new();
+        query.write(&mut buf).unwrap();
+        let msg = buf.filled();
+
+        let mut out = Vec::with_capacity(2 + msg.len());
+        out.extend_from_slice(&(msg.len() as u16).to_be_bytes());
+        out.extend_from_slice(msg);
+        stream.write_all(&out).await.unwrap();
+
+        let mut len_buf = [0u8; 2];
+        stream.read_exact(&mut len_buf).await.unwrap();
+        let resp_len = u16::from_be_bytes(len_buf) as usize;
+
+        let mut data = vec![0u8; resp_len];
+        stream.read_exact(&mut data).await.unwrap();
+
+        let mut resp_buf = BytePacketBuffer::from_bytes(&data);
+        DnsPacket::from_buffer(&mut resp_buf).unwrap()
+    }
+
+    #[tokio::test]
+    async fn dot_resolves_local_zone() {
+        let (addr, cert_der) = spawn_dot_server().await;
+        let client_config = dot_client(&cert_der, dot_alpn());
+        let mut stream = dot_connect(addr, &client_config).await;
+
+        let query = DnsPacket::query(0x1234, "dot-test.example", QueryType::A);
+        let resp = dot_exchange(&mut stream, &query).await;
+
+        assert_eq!(resp.header.id, 0x1234);
+        assert!(resp.header.response);
+        assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+        assert_eq!(resp.answers.len(), 1);
+        match &resp.answers[0] {
+            DnsRecord::A { domain, addr, ttl } => {
+                assert_eq!(domain, "dot-test.example");
+                assert_eq!(*addr, std::net::Ipv4Addr::new(10, 0, 0, 1));
+                assert_eq!(*ttl, 300);
+            }
+            other => panic!("expected A record, got {:?}", other),
+        }
+    }
+
+    #[tokio::test]
+    async fn dot_multiple_queries_on_persistent_connection() {
+        let (addr, cert_der) = spawn_dot_server().await;
+        let client_config = dot_client(&cert_der, dot_alpn());
+        let mut stream = dot_connect(addr, &client_config).await;
+
+        for i in 0..3u16 {
+            let query = DnsPacket::query(0xA000 + i, "dot-test.example", QueryType::A);
+            let resp = dot_exchange(&mut stream, &query).await;
+            assert_eq!(resp.header.id, 0xA000 + i);
+            assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+            assert_eq!(resp.answers.len(), 1);
+        }
+    }
+
+    #[tokio::test]
+    async fn dot_nxdomain_for_unknown() {
+        let (addr, cert_der) = spawn_dot_server().await;
+        let client_config = dot_client(&cert_der, dot_alpn());
+        let mut stream = dot_connect(addr, &client_config).await;
+
+        let query = DnsPacket::query(0xBEEF, "nonexistent.test", QueryType::A);
+        let resp = dot_exchange(&mut stream, &query).await;
+
+        assert_eq!(resp.header.id, 0xBEEF);
+        assert!(resp.header.response);
+        // Query goes to the blackhole upstream which never replies → SERVFAIL.
+        // The SERVFAIL response echoes the question section.
+        assert_eq!(resp.header.rescode, ResultCode::SERVFAIL);
+        assert_eq!(resp.questions.len(), 1);
+        assert_eq!(resp.questions[0].name, "nonexistent.test");
+    }
+
+    #[tokio::test]
+    async fn dot_negotiates_alpn() {
+        let (addr, cert_der) = spawn_dot_server().await;
+        let client_config = dot_client(&cert_der, dot_alpn());
+        let stream = dot_connect(addr, &client_config).await;
+        let (_io, conn) = stream.get_ref();
+        assert_eq!(conn.alpn_protocol(), Some(&b"dot"[..]));
+    }
+
+    #[tokio::test]
+    async fn dot_rejects_non_dot_alpn() {
+        // Cross-protocol confusion defense: a client that only offers "h2"
+        // (e.g. an HTTP/2 client mistakenly hitting :853) must not complete
+        // a TLS handshake with the DoT server. Verifies the rustls server
+        // sends `no_application_protocol` rather than silently negotiating.
+        let (addr, cert_der) = spawn_dot_server().await;
+        let client_config = dot_client(&cert_der, vec![b"h2".to_vec()]);
+        let connector = tokio_rustls::TlsConnector::from(client_config);
+        let tcp = tokio::net::TcpStream::connect(addr).await.unwrap();
+        let result = connector
+            .connect(ServerName::try_from("numa.numa").unwrap(), tcp)
+            .await;
+        assert!(
+            result.is_err(),
+            "DoT server must reject ALPN that doesn't include \"dot\""
+        );
+    }
+
+    #[tokio::test]
+    async fn dot_concurrent_connections() {
+        let (addr, cert_der) = spawn_dot_server().await;
+        let client_config = dot_client(&cert_der, dot_alpn());
+
+        let mut handles = Vec::new();
+        for i in 0..5u16 {
+            let cfg = Arc::clone(&client_config);
+            handles.push(tokio::spawn(async move {
+                let mut stream = dot_connect(addr, &cfg).await;
+                let query = DnsPacket::query(0xC000 + i, "dot-test.example", QueryType::A);
+                let resp = dot_exchange(&mut stream, &query).await;
+                assert_eq!(resp.header.id, 0xC000 + i);
+                assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+                assert_eq!(resp.answers.len(), 1);
+            }));
+        }
+
+        for h in handles {
+            h.await.unwrap();
+        }
+    }
+}

src/forward.rs

@@ -141,7 +141,7 @@ mod tests {
     use std::future::IntoFuture;
 
     use crate::header::ResultCode;
-    use crate::question::{DnsQuestion, QueryType};
+    use crate::question::QueryType;
     use crate::record::DnsRecord;
 
     #[test]
@@ -160,12 +160,7 @@ mod tests {
     }
 
     fn make_query() -> DnsPacket {
-        let mut q = DnsPacket::new();
-        q.header.id = 0xABCD;
-        q.header.recursion_desired = true;
-        q.questions
-            .push(DnsQuestion::new("example.com".to_string(), QueryType::A));
-        q
+        DnsPacket::query(0xABCD, "example.com", QueryType::A)
     }
 
     fn make_response(query: &DnsPacket) -> DnsPacket {

src/lib.rs

@@ -5,6 +5,7 @@ pub mod cache;
 pub mod config;
 pub mod ctx;
 pub mod dnssec;
+pub mod dot;
 pub mod forward;
 pub mod header;
 pub mod lan;
@@ -65,7 +66,9 @@ fn config_dir_unix() -> std::path::PathBuf {
     std::path::PathBuf::from("/usr/local/var/numa")
 }
 
-/// System-wide data directory for TLS certs.
+/// Default system-wide data directory for TLS certs. Overridable via
+/// `[server] data_dir = "..."` in numa.toml — this function only provides
+/// the fallback when the config doesn't set it.
 /// Unix: /usr/local/var/numa
 /// Windows: %PROGRAMDATA%\numa
 pub fn data_dir() -> std::path::PathBuf {

src/main.rs

@@ -17,10 +17,12 @@ use numa::query_log::QueryLog;
 use numa::service_store::ServiceStore;
 use numa::stats::ServerStats;
 use numa::system_dns::{
-    discover_system_dns, install_service, install_system_dns, restart_service, service_status,
-    uninstall_service, uninstall_system_dns,
+    discover_system_dns, install_service, restart_service, service_status, uninstall_service,
 };
 
+const QUAD9_IP: &str = "9.9.9.9";
+const DOH_FALLBACK: &str = "https://9.9.9.9/dns-query";
+
 #[tokio::main]
 async fn main() -> numa::Result<()> {
     env_logger::Builder::from_env(env_logger::Env::default().default_filter_or("info"))
@@ -31,12 +33,12 @@ async fn main() -> numa::Result<()> {
     let arg1 = std::env::args().nth(1).unwrap_or_default();
     match arg1.as_str() {
         "install" => {
-            eprintln!("\x1b[1;38;2;192;98;58mNuma\x1b[0m — configuring system DNS\n");
-            return install_system_dns().map_err(|e| e.into());
+            eprintln!("\x1b[1;38;2;192;98;58mNuma\x1b[0m — installing\n");
+            return install_service().map_err(|e| e.into());
         }
         "uninstall" => {
-            eprintln!("\x1b[1;38;2;192;98;58mNuma\x1b[0m — restoring system DNS\n");
-            return uninstall_system_dns().map_err(|e| e.into());
+            eprintln!("\x1b[1;38;2;192;98;58mNuma\x1b[0m — uninstalling\n");
+            return uninstall_service().map_err(|e| e.into());
         }
         "service" => {
             let sub = std::env::args().nth(2).unwrap_or_default();
@@ -107,32 +109,81 @@ async fn main() -> numa::Result<()> {
     // Discover system DNS in a single pass (upstream + forwarding rules)
     let system_dns = discover_system_dns();
 
-    let upstream_addr = if config.upstream.address.is_empty() {
-        system_dns
-            .default_upstream
-            .or_else(numa::system_dns::detect_dhcp_dns)
-            .unwrap_or_else(|| {
-                info!("could not detect system DNS, falling back to Quad9 DoH");
-                "https://dns.quad9.net/dns-query".to_string()
-            })
-    } else {
-        config.upstream.address.clone()
-    };
+    let root_hints = numa::recursive::parse_root_hints(&config.upstream.root_hints);
 
-    let upstream: Upstream = if upstream_addr.starts_with("https://") {
-        let client = reqwest::Client::builder()
-            .use_rustls_tls()
-            .build()
-            .unwrap_or_default();
-        Upstream::Doh {
-            url: upstream_addr,
-            client,
+    let (resolved_mode, upstream_auto, upstream, upstream_label) = match config.upstream.mode {
+        numa::config::UpstreamMode::Auto => {
+            info!("auto mode: probing recursive resolution...");
+            if numa::recursive::probe_recursive(&root_hints).await {
+                info!("recursive probe succeeded — self-sovereign mode");
+                let dummy = Upstream::Udp("0.0.0.0:0".parse().unwrap());
+                (
+                    numa::config::UpstreamMode::Recursive,
+                    false,
+                    dummy,
+                    "recursive (root hints)".to_string(),
+                )
+            } else {
+                log::warn!("recursive probe failed — falling back to Quad9 DoH");
+                let client = reqwest::Client::builder()
+                    .use_rustls_tls()
+                    .build()
+                    .unwrap_or_default();
+                let url = DOH_FALLBACK.to_string();
+                let label = url.clone();
+                (
+                    numa::config::UpstreamMode::Forward,
+                    false,
+                    Upstream::Doh { url, client },
+                    label,
+                )
+            }
+        }
+        numa::config::UpstreamMode::Recursive => {
+            let dummy = Upstream::Udp("0.0.0.0:0".parse().unwrap());
+            (
+                numa::config::UpstreamMode::Recursive,
+                false,
+                dummy,
+                "recursive (root hints)".to_string(),
+            )
+        }
+        numa::config::UpstreamMode::Forward => {
+            let upstream_addr = if config.upstream.address.is_empty() {
+                system_dns
+                    .default_upstream
+                    .or_else(numa::system_dns::detect_dhcp_dns)
+                    .unwrap_or_else(|| {
+                        info!("could not detect system DNS, falling back to Quad9 DoH");
+                        DOH_FALLBACK.to_string()
+                    })
+            } else {
+                config.upstream.address.clone()
+            };
+
+            let upstream: Upstream = if upstream_addr.starts_with("https://") {
+                let client = reqwest::Client::builder()
+                    .use_rustls_tls()
+                    .build()
+                    .unwrap_or_default();
+                Upstream::Doh {
+                    url: upstream_addr,
+                    client,
+                }
+            } else {
+                let addr: SocketAddr =
+                    format!("{}:{}", upstream_addr, config.upstream.port).parse()?;
+                Upstream::Udp(addr)
+            };
+            let label = upstream.to_string();
+            (
+                numa::config::UpstreamMode::Forward,
+                config.upstream.address.is_empty(),
+                upstream,
+                label,
+            )
         }
-    } else {
-        let addr: SocketAddr = format!("{}:{}", upstream_addr, config.upstream.port).parse()?;
-        Upstream::Udp(addr)
     };
-    let upstream_label = upstream.to_string();
     let api_port = config.server.api_port;
 
     let mut blocklist = BlocklistStore::new();
@@ -153,10 +204,23 @@ async fn main() -> numa::Result<()> {
 
     let forwarding_rules = system_dns.forwarding_rules;
 
+    // Resolve data_dir from config, falling back to the platform default.
+    // Used for TLS CA storage below and stored on ServerCtx for runtime use.
+    let resolved_data_dir = config
+        .server
+        .data_dir
+        .clone()
+        .unwrap_or_else(numa::data_dir);
+
     // Build initial TLS config before ServerCtx (so ArcSwap is ready at construction)
     let initial_tls = if config.proxy.enabled && config.proxy.tls_port > 0 {
         let service_names = service_store.names();
-        match numa::tls::build_tls_config(&config.proxy.tld, &service_names) {
+        match numa::tls::build_tls_config(
+            &config.proxy.tld,
+            &service_names,
+            Vec::new(),
+            &resolved_data_dir,
+        ) {
             Ok(tls_config) => Some(ArcSwap::from(tls_config)),
             Err(e) => {
                 log::warn!("TLS setup failed, HTTPS proxy disabled: {}", e);
@@ -183,7 +247,7 @@ async fn main() -> numa::Result<()> {
         lan_peers: Mutex::new(numa::lan::PeerStore::new(config.lan.peer_timeout_secs)),
         forwarding_rules,
         upstream: Mutex::new(upstream),
-        upstream_auto: config.upstream.address.is_empty(),
+        upstream_auto,
         upstream_port: config.upstream.port,
         lan_ip: Mutex::new(numa::lan::detect_lan_ip().unwrap_or(std::net::Ipv4Addr::LOCALHOST)),
         timeout: Duration::from_millis(config.upstream.timeout_ms),
@@ -197,10 +261,10 @@ async fn main() -> numa::Result<()> {
         config_path: resolved_config_path,
         config_found,
         config_dir: numa::config_dir(),
-        data_dir: numa::data_dir(),
+        data_dir: resolved_data_dir,
         tls_config: initial_tls,
-        upstream_mode: config.upstream.mode,
-        root_hints: numa::recursive::parse_root_hints(&config.upstream.root_hints),
+        upstream_mode: resolved_mode,
+        root_hints,
         srtt: std::sync::RwLock::new(numa::srtt::SrttCache::new(config.upstream.srtt)),
         inflight: std::sync::Mutex::new(std::collections::HashMap::new()),
         dnssec_enabled: config.dnssec.enabled,
@@ -208,7 +272,6 @@ async fn main() -> numa::Result<()> {
     });
 
     let zone_count: usize = ctx.zone_map.values().map(|m| m.len()).sum();
-
     // Build banner rows, then size the box to fit the longest value
     let api_url = format!("http://localhost:{}", api_port);
     let proxy_label = if config.proxy.enabled {
@@ -308,6 +371,20 @@ async fn main() -> numa::Result<()> {
     );
     if let Some(ref label) = proxy_label {
         row("Proxy", g, label);
+        if config.proxy.bind_addr == "127.0.0.1" {
+            let y = "\x1b[38;2;204;176;59m"; // yellow
+            row(
+                "",
+                y,
+                &format!(
+                    "⚠ proxy on 127.0.0.1 — .{} not LAN reachable",
+                    config.proxy.tld
+                ),
+            );
+        }
+    }
+    if config.dot.enabled {
+        row("DoT", g, &format!("tls://:{}", config.dot.port));
     }
     if config.lan.enabled {
         row("LAN", g, "mDNS (_numa._tcp.local)");
@@ -375,16 +452,11 @@ async fn main() -> numa::Result<()> {
         axum::serve(listener, app).await.unwrap();
     });
 
-    // Proxy binds 0.0.0.0 when LAN is enabled (cross-machine access), otherwise config value
-    let proxy_bind: std::net::Ipv4Addr = if config.lan.enabled {
-        std::net::Ipv4Addr::UNSPECIFIED
-    } else {
-        config
-            .proxy
-            .bind_addr
-            .parse()
-            .unwrap_or(std::net::Ipv4Addr::LOCALHOST)
-    };
+    let proxy_bind: std::net::Ipv4Addr = config
+        .proxy
+        .bind_addr
+        .parse()
+        .unwrap_or(std::net::Ipv4Addr::LOCALHOST);
 
     // Spawn HTTP reverse proxy for .numa domains
     if config.proxy.enabled {
@@ -421,11 +493,27 @@ async fn main() -> numa::Result<()> {
         });
     }
 
+    // Spawn DNS-over-TLS listener (RFC 7858)
+    if config.dot.enabled {
+        let dot_ctx = Arc::clone(&ctx);
+        let dot_config = config.dot.clone();
+        tokio::spawn(async move {
+            numa::dot::start_dot(dot_ctx, &dot_config).await;
+        });
+    }
+
     // UDP DNS listener
     #[allow(clippy::infinite_loop)]
     loop {
         let mut buffer = BytePacketBuffer::new();
-        let (_, src_addr) = ctx.socket.recv_from(&mut buffer.buf).await?;
+        let (_, src_addr) = match ctx.socket.recv_from(&mut buffer.buf).await {
+            Ok(r) => r,
+            Err(e) if e.kind() == std::io::ErrorKind::ConnectionReset => {
+                // Windows delivers ICMP port-unreachable as ConnectionReset on UDP sockets
+                continue;
+            }
+            Err(e) => return Err(e.into()),
+        };
 
         let ctx = Arc::clone(&ctx);
         tokio::spawn(async move {
@@ -468,7 +556,7 @@ async fn network_watch_loop(ctx: Arc<numa::ctx::ServerCtx>) {
             let new_addr = dns_info
                 .default_upstream
                 .or_else(numa::system_dns::detect_dhcp_dns)
-                .unwrap_or_else(|| "9.9.9.9".to_string());
+                .unwrap_or_else(|| QUAD9_IP.to_string());
             if let Ok(new_sock) =
                 format!("{}:{}", new_addr, ctx.upstream_port).parse::<SocketAddr>()
             {

src/override_store.rs

@@ -117,6 +117,22 @@ impl OverrideStore {
         self.entries.clear();
     }
 
+    pub fn heap_bytes(&self) -> usize {
+        let per_slot = std::mem::size_of::<u64>()
+            + std::mem::size_of::<String>()
+            + std::mem::size_of::<OverrideEntry>()
+            + 1;
+        let table = self.entries.capacity() * per_slot;
+        let heap: usize = self
+            .entries
+            .iter()
+            .map(|(k, v)| {
+                k.capacity() + v.domain.capacity() + v.target.capacity() + v.record.heap_bytes()
+            })
+            .sum();
+        table + heap
+    }
+
     pub fn active_count(&self) -> usize {
         self.entries.values().filter(|e| !e.is_expired()).count()
     }
@@ -154,3 +170,16 @@ fn parse_target(domain: &str, target: &str, ttl: u32) -> Result<(QueryType, DnsR
         },
     ))
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn heap_bytes_grows_with_entries() {
+        let mut store = OverrideStore::new();
+        let empty = store.heap_bytes();
+        store.insert("example.com", "1.2.3.4", 300, None).unwrap();
+        assert!(store.heap_bytes() > empty);
+    }
+}

src/packet.rs

@@ -57,6 +57,34 @@ impl DnsPacket {
         }
     }
 
+    pub fn query(id: u16, domain: &str, qtype: crate::question::QueryType) -> DnsPacket {
+        let mut pkt = DnsPacket::new();
+        pkt.header.id = id;
+        pkt.header.recursion_desired = true;
+        pkt.questions
+            .push(crate::question::DnsQuestion::new(domain.to_string(), qtype));
+        pkt
+    }
+
+    pub fn heap_bytes(&self) -> usize {
+        fn records_heap(records: &[DnsRecord]) -> usize {
+            records
+                .iter()
+                .map(|r| std::mem::size_of::<DnsRecord>() + r.heap_bytes())
+                .sum::<usize>()
+        }
+        let questions: usize = self
+            .questions
+            .iter()
+            .map(|q| std::mem::size_of::<DnsQuestion>() + q.name.capacity())
+            .sum();
+        questions
+            + records_heap(&self.answers)
+            + records_heap(&self.authorities)
+            + records_heap(&self.resources)
+            + self.edns.as_ref().map_or(0, |e| e.options.capacity())
+    }
+
     pub fn response_from(query: &DnsPacket, rescode: crate::header::ResultCode) -> DnsPacket {
         let mut resp = DnsPacket::new();
         resp.header.id = query.header.id;
@@ -582,4 +610,16 @@ mod tests {
             panic!("expected DNSKEY");
         }
     }
+
+    #[test]
+    fn heap_bytes_accounts_for_records() {
+        let mut pkt = DnsPacket::new();
+        let empty = pkt.heap_bytes();
+        pkt.answers.push(DnsRecord::A {
+            domain: "example.com".into(),
+            addr: "1.2.3.4".parse().unwrap(),
+            ttl: 300,
+        });
+        assert!(pkt.heap_bytes() > empty);
+    }
 }

src/query_log.rs

@@ -38,6 +38,21 @@ impl QueryLog {
         self.entries.push_back(entry);
     }
 
+    pub fn len(&self) -> usize {
+        self.entries.len()
+    }
+
+    pub fn is_empty(&self) -> bool {
+        self.entries.is_empty()
+    }
+
+    pub fn heap_bytes(&self) -> usize {
+        self.entries
+            .iter()
+            .map(|e| std::mem::size_of::<QueryLogEntry>() + e.domain.capacity())
+            .sum()
+    }
+
     pub fn query(&self, filter: &QueryLogFilter) -> Vec<&QueryLogEntry> {
         self.entries
             .iter()
@@ -77,3 +92,25 @@ pub struct QueryLogFilter {
     pub since: Option<SystemTime>,
     pub limit: Option<usize>,
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn heap_bytes_grows_with_entries() {
+        let mut log = QueryLog::new(100);
+        let empty = log.heap_bytes();
+        log.push(QueryLogEntry {
+            timestamp: SystemTime::now(),
+            src_addr: "127.0.0.1:1234".parse().unwrap(),
+            domain: "example.com".into(),
+            query_type: QueryType::A,
+            path: QueryPath::Forwarded,
+            rescode: ResultCode::NOERROR,
+            latency_us: 500,
+            dnssec: DnssecStatus::Indeterminate,
+        });
+        assert!(log.heap_bytes() > empty);
+    }
+}

src/record.rs

@@ -136,6 +136,46 @@ impl DnsRecord {
         }
     }
 
+    pub fn heap_bytes(&self) -> usize {
+        match self {
+            DnsRecord::A { domain, .. } => domain.capacity(),
+            DnsRecord::NS { domain, host, .. } | DnsRecord::CNAME { domain, host, .. } => {
+                domain.capacity() + host.capacity()
+            }
+            DnsRecord::MX { domain, host, .. } => domain.capacity() + host.capacity(),
+            DnsRecord::AAAA { domain, .. } => domain.capacity(),
+            DnsRecord::DNSKEY {
+                domain, public_key, ..
+            } => domain.capacity() + public_key.capacity(),
+            DnsRecord::DS { domain, digest, .. } => domain.capacity() + digest.capacity(),
+            DnsRecord::RRSIG {
+                domain,
+                signer_name,
+                signature,
+                ..
+            } => domain.capacity() + signer_name.capacity() + signature.capacity(),
+            DnsRecord::NSEC {
+                domain,
+                next_domain,
+                type_bitmap,
+                ..
+            } => domain.capacity() + next_domain.capacity() + type_bitmap.capacity(),
+            DnsRecord::NSEC3 {
+                domain,
+                salt,
+                next_hashed_owner,
+                type_bitmap,
+                ..
+            } => {
+                domain.capacity()
+                    + salt.capacity()
+                    + next_hashed_owner.capacity()
+                    + type_bitmap.capacity()
+            }
+            DnsRecord::UNKNOWN { domain, data, .. } => domain.capacity() + data.capacity(),
+        }
+    }
+
     pub fn set_ttl(&mut self, new_ttl: u32) {
         match self {
             DnsRecord::A { ttl, .. }
@@ -650,4 +690,14 @@ mod tests {
         let parsed = round_trip(&rec);
         assert_eq!(rec, parsed);
     }
+
+    #[test]
+    fn heap_bytes_reflects_string_capacity() {
+        let rec = DnsRecord::CNAME {
+            domain: "a]".repeat(100),
+            host: "b".repeat(200),
+            ttl: 60,
+        };
+        assert!(rec.heap_bytes() >= 300);
+    }
 }

src/recursive.rs

@@ -9,7 +9,7 @@ use crate::cache::DnsCache;
 use crate::forward::forward_udp;
 use crate::header::ResultCode;
 use crate::packet::DnsPacket;
-use crate::question::{DnsQuestion, QueryType};
+use crate::question::QueryType;
 use crate::record::DnsRecord;
 use crate::srtt::SrttCache;
 
@@ -32,6 +32,14 @@ fn dns_addr(ip: impl Into<IpAddr>) -> SocketAddr {
     SocketAddr::new(ip.into(), 53)
 }
 
+fn record_to_addr(rec: &DnsRecord) -> Option<SocketAddr> {
+    match rec {
+        DnsRecord::A { addr, .. } => Some(dns_addr(*addr)),
+        DnsRecord::AAAA { addr, .. } => Some(dns_addr(*addr)),
+        _ => None,
+    }
+}
+
 pub fn reset_udp_state() {
     UDP_DISABLED.store(false, Ordering::Release);
     UDP_FAILURES.store(0, Ordering::Release);
@@ -46,11 +54,8 @@ pub async fn probe_udp(root_hints: &[SocketAddr]) {
         Some(h) => *h,
         None => return,
     };
-    let mut probe = DnsPacket::new();
-    probe.header.id = next_id();
-    probe
-        .questions
-        .push(DnsQuestion::new(".".to_string(), QueryType::NS));
+    let mut probe = DnsPacket::query(next_id(), ".", QueryType::NS);
+    probe.header.recursion_desired = false;
     if forward_udp(&probe, hint, Duration::from_millis(1500))
         .await
         .is_ok()
@@ -60,6 +65,21 @@ pub async fn probe_udp(root_hints: &[SocketAddr]) {
     }
 }
 
+/// Probe whether recursive resolution works by querying root servers.
+/// Tries up to 3 hints before declaring failure.
+pub async fn probe_recursive(root_hints: &[SocketAddr]) -> bool {
+    let mut probe = DnsPacket::query(next_id(), ".", QueryType::NS);
+    probe.header.recursion_desired = false;
+    for hint in root_hints.iter().take(3) {
+        if let Ok(resp) = forward_udp(&probe, *hint, Duration::from_secs(3)).await {
+            if !resp.answers.is_empty() || !resp.authorities.is_empty() {
+                return true;
+            }
+        }
+    }
+    false
+}
+
 pub async fn prime_tld_cache(
     cache: &RwLock<DnsCache>,
     root_hints: &[SocketAddr],
@@ -296,17 +316,8 @@ pub(crate) fn resolve_iterative<'a>(
                             )
                             .await
                             {
-                                for rec in &ns_resp.answers {
-                                    match rec {
-                                        DnsRecord::A { addr, .. } => {
-                                            new_ns_addrs.push(dns_addr(*addr));
-                                        }
-                                        DnsRecord::AAAA { addr, .. } => {
-                                            new_ns_addrs.push(dns_addr(*addr));
-                                        }
-                                        _ => {}
-                                    }
-                                }
+                                new_ns_addrs
+                                    .extend(ns_resp.answers.iter().filter_map(record_to_addr));
                             }
                             if !new_ns_addrs.is_empty() {
                                 break;
@@ -360,13 +371,7 @@ fn find_closest_ns(
                 if let DnsRecord::NS { host, .. } = ns_rec {
                     for qt in [QueryType::A, QueryType::AAAA] {
                         if let Some(resp) = guard.lookup(host, qt) {
-                            for rec in &resp.answers {
-                                match rec {
-                                    DnsRecord::A { addr, .. } => addrs.push(dns_addr(*addr)),
-                                    DnsRecord::AAAA { addr, .. } => addrs.push(dns_addr(*addr)),
-                                    _ => {}
-                                }
-                            }
+                            addrs.extend(resp.answers.iter().filter_map(record_to_addr));
                         }
                     }
                 }
@@ -452,13 +457,7 @@ fn addrs_from_cache(cache: &RwLock<DnsCache>, name: &str) -> Vec<SocketAddr> {
     let mut addrs = Vec::new();
     for qt in [QueryType::A, QueryType::AAAA] {
         if let Some(pkt) = guard.lookup(name, qt) {
-            for rec in &pkt.answers {
-                match rec {
-                    DnsRecord::A { addr, .. } => addrs.push(dns_addr(*addr)),
-                    DnsRecord::AAAA { addr, .. } => addrs.push(dns_addr(*addr)),
-                    _ => {}
-                }
-            }
+            addrs.extend(pkt.answers.iter().filter_map(record_to_addr));
         }
     }
     addrs
@@ -468,15 +467,13 @@ fn glue_addrs_for(response: &DnsPacket, ns_name: &str) -> Vec<SocketAddr> {
     response
         .resources
         .iter()
-        .filter_map(|r| match r {
-            DnsRecord::A { domain, addr, .. } if domain.eq_ignore_ascii_case(ns_name) => {
-                Some(dns_addr(*addr))
+        .filter(|r| match r {
+            DnsRecord::A { domain, .. } | DnsRecord::AAAA { domain, .. } => {
+                domain.eq_ignore_ascii_case(ns_name)
             }
-            DnsRecord::AAAA { domain, addr, .. } if domain.eq_ignore_ascii_case(ns_name) => {
-                Some(dns_addr(*addr))
-            }
-            _ => None,
+            _ => false,
         })
+        .filter_map(record_to_addr)
         .collect()
 }
 
@@ -596,12 +593,8 @@ async fn send_query(
     server: SocketAddr,
     srtt: &RwLock<SrttCache>,
 ) -> crate::Result<DnsPacket> {
-    let mut query = DnsPacket::new();
-    query.header.id = next_id();
+    let mut query = DnsPacket::query(next_id(), qname, qtype);
     query.header.recursion_desired = false;
-    query
-        .questions
-        .push(DnsQuestion::new(qname.to_string(), qtype));
     query.edns = Some(crate::packet::EdnsOpt {
         do_bit: true,
         ..Default::default()
@@ -877,14 +870,25 @@ mod tests {
                 };
                 let handler = handler.clone();
                 tokio::spawn(async move {
+                    let timeout = std::time::Duration::from_secs(5);
                     // Read length-prefixed DNS query
                     let mut len_buf = [0u8; 2];
-                    if stream.read_exact(&mut len_buf).await.is_err() {
+                    if tokio::time::timeout(timeout, stream.read_exact(&mut len_buf))
+                        .await
+                        .ok()
+                        .and_then(|r| r.ok())
+                        .is_none()
+                    {
                         return;
                     }
                     let len = u16::from_be_bytes(len_buf) as usize;
                     let mut data = vec![0u8; len];
-                    if stream.read_exact(&mut data).await.is_err() {
+                    if tokio::time::timeout(timeout, stream.read_exact(&mut data))
+                        .await
+                        .ok()
+                        .and_then(|r| r.ok())
+                        .is_none()
+                    {
                         return;
                     }
 
@@ -1056,11 +1060,7 @@ mod tests {
         })
         .await;
 
-        let mut query = DnsPacket::new();
-        query.header.id = 0xBEEF;
-        query
-            .questions
-            .push(DnsQuestion::new("test.com".to_string(), QueryType::A));
+        let query = DnsPacket::query(0xBEEF, "test.com", QueryType::A);
 
         let resp = crate::forward::forward_tcp(&query, server_addr, Duration::from_secs(2))
             .await
@@ -1120,11 +1120,7 @@ mod tests {
                 .unwrap();
         });
 
-        let mut query = DnsPacket::new();
-        query.header.id = 0xCAFE;
-        query
-            .questions
-            .push(DnsQuestion::new("strict.test".to_string(), QueryType::A));
+        let query = DnsPacket::query(0xCAFE, "strict.test", QueryType::A);
 
         let resp = crate::forward::forward_tcp(&query, addr, Duration::from_secs(2))
             .await

src/srtt.rs

@@ -47,16 +47,19 @@ impl SrttCache {
 
     /// Apply time-based decay: each DECAY_AFTER_SECS period halves distance to INITIAL.
     fn decayed_srtt(entry: &SrttEntry) -> u64 {
-        let age_secs = entry.updated_at.elapsed().as_secs();
+        Self::decay_for_age(entry.srtt_ms, entry.updated_at.elapsed().as_secs())
+    }
+
+    fn decay_for_age(srtt_ms: u64, age_secs: u64) -> u64 {
         if age_secs > DECAY_AFTER_SECS {
             let periods = (age_secs / DECAY_AFTER_SECS).min(8);
-            let mut srtt = entry.srtt_ms;
+            let mut srtt = srtt_ms;
             for _ in 0..periods {
                 srtt = (srtt + INITIAL_SRTT_MS) / 2;
             }
             srtt
         } else {
-            entry.srtt_ms
+            srtt_ms
         }
     }
 
@@ -100,6 +103,14 @@ impl SrttCache {
         addrs.sort_by_key(|a| self.get(a.ip()));
     }
 
+    pub fn heap_bytes(&self) -> usize {
+        let per_slot = std::mem::size_of::<u64>()
+            + std::mem::size_of::<IpAddr>()
+            + std::mem::size_of::<SrttEntry>()
+            + 1;
+        self.entries.capacity() * per_slot
+    }
+
     pub fn len(&self) -> usize {
         self.entries.len()
     }
@@ -108,13 +119,6 @@ impl SrttCache {
         self.entries.is_empty()
     }
 
-    #[cfg(test)]
-    fn set_updated_at(&mut self, ip: IpAddr, at: Instant) {
-        if let Some(entry) = self.entries.get_mut(&ip) {
-            entry.updated_at = at;
-        }
-    }
-
     fn maybe_evict(&mut self) {
         if self.entries.len() < MAX_ENTRIES {
             return;
@@ -210,63 +214,41 @@ mod tests {
         assert_eq!(addrs, original);
     }
 
-    fn age(secs: u64) -> Instant {
-        Instant::now() - std::time::Duration::from_secs(secs)
-    }
-
-    /// Cache with ip(1) saturated at FAILURE_PENALTY_MS
-    fn saturated_penalty_cache() -> SrttCache {
-        let mut cache = SrttCache::new(true);
-        for _ in 0..30 {
-            cache.record_rtt(ip(1), FAILURE_PENALTY_MS, false);
-        }
-        cache
-    }
-
     #[test]
     fn no_decay_within_threshold() {
-        let mut cache = SrttCache::new(true);
-        cache.record_rtt(ip(1), 5000, false);
-        cache.set_updated_at(ip(1), age(DECAY_AFTER_SECS));
-        assert_eq!(cache.get(ip(1)), cache.entries[&ip(1)].srtt_ms);
+        // At exactly DECAY_AFTER_SECS, no decay applied
+        let result = SrttCache::decay_for_age(FAILURE_PENALTY_MS, DECAY_AFTER_SECS);
+        assert_eq!(result, FAILURE_PENALTY_MS);
     }
 
     #[test]
     fn one_decay_period() {
-        let mut cache = saturated_penalty_cache();
-        let raw = cache.entries[&ip(1)].srtt_ms;
-        cache.set_updated_at(ip(1), age(DECAY_AFTER_SECS + 1));
-        let expected = (raw + INITIAL_SRTT_MS) / 2;
-        assert_eq!(cache.get(ip(1)), expected);
+        let result = SrttCache::decay_for_age(FAILURE_PENALTY_MS, DECAY_AFTER_SECS + 1);
+        let expected = (FAILURE_PENALTY_MS + INITIAL_SRTT_MS) / 2;
+        assert_eq!(result, expected);
     }
 
     #[test]
     fn multiple_decay_periods() {
-        let mut cache = saturated_penalty_cache();
-        let raw = cache.entries[&ip(1)].srtt_ms;
-        cache.set_updated_at(ip(1), age(DECAY_AFTER_SECS * 4 + 1));
-        let mut expected = raw;
+        let result = SrttCache::decay_for_age(FAILURE_PENALTY_MS, DECAY_AFTER_SECS * 4 + 1);
+        let mut expected = FAILURE_PENALTY_MS;
         for _ in 0..4 {
             expected = (expected + INITIAL_SRTT_MS) / 2;
         }
-        assert_eq!(cache.get(ip(1)), expected);
+        assert_eq!(result, expected);
     }
 
     #[test]
     fn decay_caps_at_8_periods() {
         // 9 periods and 100 periods should produce the same result (capped at 8)
-        let mut cache_a = saturated_penalty_cache();
-        let mut cache_b = saturated_penalty_cache();
-        cache_a.set_updated_at(ip(1), age(DECAY_AFTER_SECS * 9 + 1));
-        cache_b.set_updated_at(ip(1), age(DECAY_AFTER_SECS * 100));
-        assert_eq!(cache_a.get(ip(1)), cache_b.get(ip(1)));
+        let a = SrttCache::decay_for_age(FAILURE_PENALTY_MS, DECAY_AFTER_SECS * 9 + 1);
+        let b = SrttCache::decay_for_age(FAILURE_PENALTY_MS, DECAY_AFTER_SECS * 100);
+        assert_eq!(a, b);
     }
 
     #[test]
     fn decay_converges_toward_initial() {
-        let mut cache = saturated_penalty_cache();
-        cache.set_updated_at(ip(1), age(DECAY_AFTER_SECS * 100));
-        let decayed = cache.get(ip(1));
+        let decayed = SrttCache::decay_for_age(FAILURE_PENALTY_MS, DECAY_AFTER_SECS * 100);
         let diff = decayed.abs_diff(INITIAL_SRTT_MS);
         assert!(
             diff < 25,
@@ -278,29 +260,38 @@ mod tests {
 
     #[test]
     fn record_rtt_applies_decay_before_ewma() {
-        let mut cache = saturated_penalty_cache();
-        cache.set_updated_at(ip(1), age(DECAY_AFTER_SECS * 8));
-        cache.record_rtt(ip(1), 50, false);
-        let srtt = cache.get(ip(1));
-        // Without decay-before-EWMA, result would be ~(5000*7+50)/8 ≈ 4381
-        assert!(srtt < 500, "expected decay before EWMA, got srtt={}", srtt);
+        // Verify decay is applied before EWMA in record_rtt by checking
+        // that a saturated penalty + long age + new sample produces a low SRTT
+        let decayed = SrttCache::decay_for_age(FAILURE_PENALTY_MS, DECAY_AFTER_SECS * 8);
+        // EWMA: (decayed * 7 + 50) / 8
+        let after_ewma = (decayed * 7 + 50) / 8;
+        assert!(
+            after_ewma < 500,
+            "expected decay before EWMA, got srtt={}",
+            after_ewma
+        );
     }
 
     #[test]
     fn decay_reranks_stale_failures() {
-        let mut cache = saturated_penalty_cache();
-        for _ in 0..30 {
-            cache.record_rtt(ip(2), 300, false);
-        }
-        let mut addrs = vec![sock(1), sock(2)];
-        cache.sort_by_rtt(&mut addrs);
-        assert_eq!(addrs, vec![sock(2), sock(1)]);
+        // After enough decay, a failed server (5000ms) converges toward
+        // INITIAL (200ms), which is below a stable server at 300ms
+        let decayed = SrttCache::decay_for_age(FAILURE_PENALTY_MS, DECAY_AFTER_SECS * 100);
+        assert!(
+            decayed < 300,
+            "expected decayed penalty ({}) < 300ms",
+            decayed
+        );
+    }
 
-        // Age server 1 so it decays toward INITIAL (200ms) — below server 2's 300ms
-        cache.set_updated_at(ip(1), age(DECAY_AFTER_SECS * 100));
-        let mut addrs = vec![sock(1), sock(2)];
-        cache.sort_by_rtt(&mut addrs);
-        assert_eq!(addrs, vec![sock(1), sock(2)]);
+    #[test]
+    fn heap_bytes_grows_with_entries() {
+        let mut cache = SrttCache::new(true);
+        let empty = cache.heap_bytes();
+        for i in 1..=10u8 {
+            cache.record_rtt(ip(i), 100, false);
+        }
+        assert!(cache.heap_bytes() > empty);
     }
 
     #[test]

src/stats.rs

@@ -1,5 +1,92 @@
 use std::time::Instant;
 
+/// Returns the process memory footprint in bytes, or 0 if unavailable.
+/// macOS: phys_footprint (matches Activity Monitor). Linux: RSS from /proc/self/statm.
+pub fn process_memory_bytes() -> usize {
+    #[cfg(target_os = "macos")]
+    {
+        macos_rss()
+    }
+    #[cfg(target_os = "linux")]
+    {
+        linux_rss()
+    }
+    #[cfg(not(any(target_os = "macos", target_os = "linux")))]
+    {
+        0
+    }
+}
+
+#[cfg(target_os = "macos")]
+fn macos_rss() -> usize {
+    use std::mem;
+    extern "C" {
+        fn mach_task_self() -> u32;
+        fn task_info(
+            target_task: u32,
+            flavor: u32,
+            task_info_out: *mut TaskVmInfo,
+            task_info_count: *mut u32,
+        ) -> i32;
+    }
+    // Partial task_vm_info_data_t — only fields up to phys_footprint.
+    #[repr(C)]
+    struct TaskVmInfo {
+        virtual_size: u64,
+        region_count: i32,
+        page_size: i32,
+        resident_size: u64,
+        resident_size_peak: u64,
+        device: u64,
+        device_peak: u64,
+        internal: u64,
+        internal_peak: u64,
+        external: u64,
+        external_peak: u64,
+        reusable: u64,
+        reusable_peak: u64,
+        purgeable_volatile_pmap: u64,
+        purgeable_volatile_resident: u64,
+        purgeable_volatile_virtual: u64,
+        compressed: u64,
+        compressed_peak: u64,
+        compressed_lifetime: u64,
+        phys_footprint: u64,
+    }
+    const TASK_VM_INFO: u32 = 22;
+    let mut info: TaskVmInfo = unsafe { mem::zeroed() };
+    let mut count = (mem::size_of::<TaskVmInfo>() / mem::size_of::<u32>()) as u32;
+    let kr = unsafe { task_info(mach_task_self(), TASK_VM_INFO, &mut info, &mut count) };
+    if kr == 0 {
+        info.phys_footprint as usize
+    } else {
+        0
+    }
+}
+
+#[cfg(target_os = "linux")]
+fn linux_rss() -> usize {
+    extern "C" {
+        fn sysconf(name: i32) -> i64;
+    }
+    const SC_PAGESIZE: i32 = 30; // x86_64 + aarch64; differs on mips (28), sparc (29)
+    let page_size = unsafe { sysconf(SC_PAGESIZE) };
+    let page_size = if page_size > 0 {
+        page_size as usize
+    } else {
+        4096
+    };
+
+    if let Ok(statm) = std::fs::read_to_string("/proc/self/statm") {
+        if let Some(rss_pages) = statm.split_whitespace().nth(1) {
+            if let Ok(pages) = rss_pages.parse::<usize>() {
+                return pages * page_size;
+            }
+        }
+    }
+    0
+}
+
 pub struct ServerStats {
     queries_total: u64,
     queries_forwarded: u64,
@@ -13,7 +100,7 @@ pub struct ServerStats {
     started_at: Instant,
 }
 
-#[derive(Clone, Copy, PartialEq, Eq)]
+#[derive(Clone, Copy, Debug, PartialEq, Eq)]
 pub enum QueryPath {
     Local,
     Cached,

src/system_dns.rs

@@ -2,6 +2,10 @@ use std::net::SocketAddr;
 
 use log::info;
 
+fn is_loopback_or_stub(addr: &str) -> bool {
+    matches!(addr, "127.0.0.1" | "127.0.0.53" | "0.0.0.0" | "::1" | "")
+}
+
 /// A conditional forwarding rule: domains matching `suffix` are forwarded to `upstream`.
 #[derive(Debug, Clone)]
 pub struct ForwardingRule {
@@ -26,10 +30,7 @@ pub fn discover_system_dns() -> SystemDnsInfo {
     }
     #[cfg(target_os = "linux")]
     {
-        SystemDnsInfo {
-            default_upstream: detect_upstream_linux_or_backup(),
-            forwarding_rules: Vec::new(),
-        }
+        discover_linux()
     }
     #[cfg(windows)]
     {
@@ -102,11 +103,7 @@ fn discover_macos() -> SystemDnsInfo {
                 if ns.parse::<std::net::Ipv4Addr>().is_ok() {
                     current_nameserver = Some(ns.clone());
                     // Capture first non-supplemental, non-loopback nameserver as default upstream
-                    if !is_supplemental
-                        && default_upstream.is_none()
-                        && ns != "127.0.0.1"
-                        && ns != "0.0.0.0"
-                    {
+                    if !is_supplemental && default_upstream.is_none() && !is_loopback_or_stub(&ns) {
                         default_upstream = Some(ns);
                     }
                 }
@@ -156,7 +153,7 @@ fn discover_macos() -> SystemDnsInfo {
     }
 }
 
-#[cfg(target_os = "macos")]
+#[cfg(any(target_os = "macos", target_os = "linux"))]
 fn make_rule(domain: &str, nameserver: &str) -> Option<ForwardingRule> {
     let addr: SocketAddr = format!("{}:53", nameserver).parse().ok()?;
     Some(ForwardingRule {
@@ -166,38 +163,100 @@ fn make_rule(domain: &str, nameserver: &str) -> Option<ForwardingRule> {
     })
 }
 
-/// Detect upstream from /etc/resolv.conf, falling back to backup file if resolv.conf
-/// only has loopback (meaning numa install already ran).
 #[cfg(target_os = "linux")]
-fn detect_upstream_linux_or_backup() -> Option<String> {
-    // Try /etc/resolv.conf first
-    if let Some(ns) = read_upstream_from_file("/etc/resolv.conf") {
-        info!("detected system upstream: {}", ns);
-        return Some(ns);
-    }
-    // If resolv.conf only has loopback, check the backup from `numa install`
-    let backup = {
-        let home = std::env::var("HOME")
-            .map(std::path::PathBuf::from)
-            .unwrap_or_else(|_| std::path::PathBuf::from("/root"));
-        home.join(".numa").join("original-resolv.conf")
-    };
-    if let Some(ns) = read_upstream_from_file(backup.to_str().unwrap_or("")) {
-        info!("detected original upstream from backup: {}", ns);
-        return Some(ns);
-    }
-    None
-}
+const CLOUD_VPC_RESOLVER: &str = "169.254.169.253";
 
 #[cfg(target_os = "linux")]
-fn read_upstream_from_file(path: &str) -> Option<String> {
-    let text = std::fs::read_to_string(path).ok()?;
+fn discover_linux() -> SystemDnsInfo {
+    // Parse resolv.conf once for both upstream and search domains
+    let (upstream, search_domains) = parse_resolv_conf("/etc/resolv.conf");
+
+    let default_upstream = if let Some(ns) = upstream {
+        info!("detected system upstream: {}", ns);
+        Some(ns)
+    } else {
+        // Fallback to backup from a previous `numa install`
+        let backup = {
+            let home = std::env::var("HOME")
+                .map(std::path::PathBuf::from)
+                .unwrap_or_else(|_| std::path::PathBuf::from("/root"));
+            home.join(".numa").join("original-resolv.conf")
+        };
+        let (ns, _) = parse_resolv_conf(backup.to_str().unwrap_or(""));
+        if let Some(ref ns) = ns {
+            info!("detected original upstream from backup: {}", ns);
+        }
+        ns
+    };
+
+    // On cloud VMs (AWS/GCP), internal domains need to reach the VPC resolver
+    let forwarding_rules = if search_domains.is_empty() {
+        Vec::new()
+    } else {
+        let forwarder = resolvectl_dns_server().unwrap_or_else(|| CLOUD_VPC_RESOLVER.to_string());
+        let rules: Vec<_> = search_domains
+            .iter()
+            .filter_map(|domain| {
+                let rule = make_rule(domain, &forwarder)?;
+                info!("forwarding .{} to {}", domain, forwarder);
+                Some(rule)
+            })
+            .collect();
+        if !rules.is_empty() {
+            info!("detected {} search domain forwarding rules", rules.len());
+        }
+        rules
+    };
+
+    SystemDnsInfo {
+        default_upstream,
+        forwarding_rules,
+    }
+}
+
+/// Parse resolv.conf in a single pass, extracting both the first non-loopback
+/// nameserver and all search domains.
+#[cfg(target_os = "linux")]
+fn parse_resolv_conf(path: &str) -> (Option<String>, Vec<String>) {
+    let text = match std::fs::read_to_string(path) {
+        Ok(t) => t,
+        Err(_) => return (None, Vec::new()),
+    };
+    let mut upstream = None;
+    let mut search_domains = Vec::new();
     for line in text.lines() {
         let line = line.trim();
         if line.starts_with("nameserver") {
-            if let Some(ns) = line.split_whitespace().nth(1) {
-                if ns != "127.0.0.1" && ns != "0.0.0.0" && ns != "::1" {
-                    return Some(ns.to_string());
+            if upstream.is_none() {
+                if let Some(ns) = line.split_whitespace().nth(1) {
+                    if !is_loopback_or_stub(ns) {
+                        upstream = Some(ns.to_string());
+                    }
+                }
+            }
+        } else if line.starts_with("search") || line.starts_with("domain") {
+            for domain in line.split_whitespace().skip(1) {
+                search_domains.push(domain.to_string());
+            }
+        }
+    }
+    (upstream, search_domains)
+}
+
+/// Query resolvectl for the real upstream DNS server (e.g. VPC resolver on AWS).
+#[cfg(target_os = "linux")]
+fn resolvectl_dns_server() -> Option<String> {
+    let output = std::process::Command::new("resolvectl")
+        .args(["status", "--no-pager"])
+        .output()
+        .ok()?;
+    let text = String::from_utf8_lossy(&output.stdout);
+    for line in text.lines() {
+        if line.contains("DNS Servers") || line.contains("Current DNS Server") {
+            if let Some(ip) = line.split(':').next_back() {
+                let ip = ip.trim();
+                if ip.parse::<std::net::IpAddr>().is_ok() && !is_loopback_or_stub(ip) {
+                    return Some(ip.to_string());
                 }
             }
         }
@@ -236,10 +295,7 @@ fn detect_dhcp_dns_macos() -> Option<String> {
                     // Take the first non-loopback DNS server
                     for addr in inner.split(',') {
                         let addr = addr.trim();
-                        if !addr.is_empty()
-                            && addr != "127.0.0.1"
-                            && addr != "0.0.0.0"
-                            && addr.parse::<std::net::Ipv4Addr>().is_ok()
+                        if !is_loopback_or_stub(addr) && addr.parse::<std::net::Ipv4Addr>().is_ok()
                         {
                             log::info!("detected DHCP DNS: {}", addr);
                             return Some(addr.to_string());
@@ -278,7 +334,7 @@ fn discover_windows() -> SystemDnsInfo {
         if trimmed.contains("DNS Servers") || trimmed.contains("DNS-Server") {
             if let Some(ip) = trimmed.split(':').next_back() {
                 let ip = ip.trim();
-                if !ip.is_empty() && ip != "127.0.0.1" && ip != "::1" {
+                if ip.parse::<std::net::IpAddr>().is_ok() && !is_loopback_or_stub(ip) {
                     upstream = Some(ip.to_string());
                     break;
                 }
@@ -302,6 +358,339 @@ fn discover_windows() -> SystemDnsInfo {
     }
 }
 
+#[cfg(any(windows, test))]
+#[derive(serde::Serialize, serde::Deserialize, Debug, PartialEq)]
+struct WindowsInterfaceDns {
+    dhcp: bool,
+    servers: Vec<String>,
+}
+
+#[cfg(any(windows, test))]
+fn parse_ipconfig_interfaces(text: &str) -> std::collections::HashMap<String, WindowsInterfaceDns> {
+    let mut interfaces = std::collections::HashMap::new();
+    let mut current_adapter: Option<String> = None;
+    let mut current_dhcp = false;
+    let mut current_dns: Vec<String> = Vec::new();
+    let mut in_dns_block = false;
+    let mut disconnected = false;
+
+    for line in text.lines() {
+        let trimmed = line.trim();
+
+        // Adapter section headers start at column 0
+        if !trimmed.is_empty() && !line.starts_with(' ') && !line.starts_with('\t') {
+            if let Some(name) = current_adapter.take() {
+                if !disconnected {
+                    interfaces.insert(
+                        name,
+                        WindowsInterfaceDns {
+                            dhcp: current_dhcp,
+                            servers: std::mem::take(&mut current_dns),
+                        },
+                    );
+                }
+                current_dns.clear();
+            }
+            in_dns_block = false;
+            current_dhcp = false;
+            disconnected = false;
+
+            // "XXX adapter YYY:" (English) / "XXX Adapter YYY:" (German)
+            let lower = trimmed.to_lowercase();
+            if let Some(pos) = lower.find(" adapter ") {
+                let after = &trimmed[pos + " adapter ".len()..];
+                let name = after.trim_end_matches(':').trim();
+                if !name.is_empty() {
+                    current_adapter = Some(name.to_string());
+                }
+            }
+        } else if current_adapter.is_some() {
+            if trimmed.contains("Media disconnected") || trimmed.contains("Medienstatus") {
+                disconnected = true;
+            } else if trimmed.contains("DHCP") && trimmed.contains(". .") {
+                current_dhcp = trimmed
+                    .split(':')
+                    .next_back()
+                    .map(|v| {
+                        let v = v.trim().to_lowercase();
+                        v == "yes" || v == "ja"
+                    })
+                    .unwrap_or(false);
+                in_dns_block = false;
+            } else if trimmed.contains("DNS Servers") || trimmed.contains("DNS-Server") {
+                in_dns_block = true;
+                if let Some(ip) = trimmed.split(':').next_back() {
+                    let ip = ip.trim();
+                    if ip.parse::<std::net::IpAddr>().is_ok() {
+                        current_dns.push(ip.to_string());
+                    }
+                }
+            } else if in_dns_block {
+                if trimmed.parse::<std::net::IpAddr>().is_ok() {
+                    current_dns.push(trimmed.to_string());
+                } else {
+                    in_dns_block = false;
+                }
+            }
+        }
+    }
+
+    if let Some(name) = current_adapter {
+        if !disconnected {
+            interfaces.insert(
+                name,
+                WindowsInterfaceDns {
+                    dhcp: current_dhcp,
+                    servers: current_dns,
+                },
+            );
+        }
+    }
+
+    interfaces
+}
+
+#[cfg(windows)]
+fn get_windows_interfaces() -> Result<std::collections::HashMap<String, WindowsInterfaceDns>, String>
+{
+    let output = std::process::Command::new("ipconfig")
+        .arg("/all")
+        .output()
+        .map_err(|e| format!("failed to run ipconfig /all: {}", e))?;
+    let text = String::from_utf8_lossy(&output.stdout);
+    Ok(parse_ipconfig_interfaces(&text))
+}
+
+#[cfg(windows)]
+fn windows_backup_path() -> std::path::PathBuf {
+    // Use ProgramData (not APPDATA) since install requires admin elevation
+    // and APPDATA differs between user and admin contexts.
+    std::path::PathBuf::from(
+        std::env::var("PROGRAMDATA").unwrap_or_else(|_| "C:\\ProgramData".into()),
+    )
+    .join("numa")
+    .join("original-dns.json")
+}
+
+#[cfg(windows)]
+fn disable_dnscache() -> Result<bool, String> {
+    // Check if Dnscache is running (it holds port 53 at kernel level)
+    let output = std::process::Command::new("sc")
+        .args(["query", "Dnscache"])
+        .output()
+        .map_err(|e| format!("failed to query Dnscache: {}", e))?;
+    let text = String::from_utf8_lossy(&output.stdout);
+    if !text.contains("RUNNING") {
+        return Ok(false);
+    }
+
+    eprintln!("  Disabling DNS Client (Dnscache) to free port 53...");
+    // Dnscache can't be stopped via sc/net stop — must disable via registry
+    let status = std::process::Command::new("reg")
+        .args([
+            "add",
+            "HKLM\\SYSTEM\\CurrentControlSet\\Services\\Dnscache",
+            "/v",
+            "Start",
+            "/t",
+            "REG_DWORD",
+            "/d",
+            "4",
+            "/f",
+        ])
+        .status()
+        .map_err(|e| format!("failed to disable Dnscache: {}", e))?;
+
+    if !status.success() {
+        return Err("failed to disable Dnscache via registry (run as Administrator?)".into());
+    }
+
+    eprintln!("  Dnscache disabled. A reboot is required to free port 53.");
+    Ok(true)
+}
+
+#[cfg(windows)]
+fn enable_dnscache() {
+    let _ = std::process::Command::new("reg")
+        .args([
+            "add",
+            "HKLM\\SYSTEM\\CurrentControlSet\\Services\\Dnscache",
+            "/v",
+            "Start",
+            "/t",
+            "REG_DWORD",
+            "/d",
+            "2",
+            "/f",
+        ])
+        .status();
+}
+
+#[cfg(windows)]
+fn install_windows() -> Result<(), String> {
+    let interfaces = get_windows_interfaces()?;
+    if interfaces.is_empty() {
+        return Err("no active network interfaces found".to_string());
+    }
+
+    let path = windows_backup_path();
+    if let Some(parent) = path.parent() {
+        std::fs::create_dir_all(parent)
+            .map_err(|e| format!("failed to create {}: {}", parent.display(), e))?;
+    }
+    let json = serde_json::to_string_pretty(&interfaces)
+        .map_err(|e| format!("failed to serialize backup: {}", e))?;
+    std::fs::write(&path, json).map_err(|e| format!("failed to write backup: {}", e))?;
+
+    for name in interfaces.keys() {
+        let status = std::process::Command::new("netsh")
+            .args([
+                "interface",
+                "ipv4",
+                "set",
+                "dnsservers",
+                name,
+                "static",
+                "127.0.0.1",
+                "primary",
+            ])
+            .status()
+            .map_err(|e| format!("failed to set DNS for {}: {}", name, e))?;
+
+        if status.success() {
+            eprintln!("  set DNS for \"{}\" -> 127.0.0.1", name);
+        } else {
+            eprintln!(
+                "  warning: failed to set DNS for \"{}\" (run as Administrator?)",
+                name
+            );
+        }
+    }
+
+    let needs_reboot = disable_dnscache()?;
+    register_autostart();
+
+    eprintln!("\n  Original DNS saved to {}", path.display());
+    eprintln!("  Run 'numa uninstall' to restore.\n");
+    if needs_reboot {
+        eprintln!("  *** Reboot required. Numa will start automatically. ***\n");
+    } else {
+        eprintln!("  Numa will start automatically on next boot.\n");
+    }
+    eprintln!("  Want full DNS sovereignty? Add to numa.toml:");
+    eprintln!("    [upstream]");
+    eprintln!("    mode = \"recursive\"\n");
+    Ok(())
+}
+
+/// Register numa to auto-start on boot via registry Run key.
+#[cfg(windows)]
+fn register_autostart() {
+    let exe = std::env::current_exe()
+        .map(|p| p.to_string_lossy().to_string())
+        .unwrap_or_else(|_| "numa".into());
+    let _ = std::process::Command::new("reg")
+        .args([
+            "add",
+            "HKLM\\SOFTWARE\\Microsoft\\Windows\\CurrentVersion\\Run",
+            "/v",
+            "Numa",
+            "/t",
+            "REG_SZ",
+            "/d",
+            &exe,
+            "/f",
+        ])
+        .status();
+    eprintln!("  Registered auto-start on boot.");
+}
+
+/// Remove numa auto-start registry key.
+#[cfg(windows)]
+fn remove_autostart() {
+    let _ = std::process::Command::new("reg")
+        .args([
+            "delete",
+            "HKLM\\SOFTWARE\\Microsoft\\Windows\\CurrentVersion\\Run",
+            "/v",
+            "Numa",
+            "/f",
+        ])
+        .status();
+}
+
+#[cfg(windows)]
+fn uninstall_windows() -> Result<(), String> {
+    remove_autostart();
+    let path = windows_backup_path();
+    let json = std::fs::read_to_string(&path)
+        .map_err(|e| format!("no backup found at {}: {}", path.display(), e))?;
+    let original: std::collections::HashMap<String, WindowsInterfaceDns> =
+        serde_json::from_str(&json).map_err(|e| format!("invalid backup file: {}", e))?;
+
+    for (name, dns_info) in &original {
+        if dns_info.dhcp || dns_info.servers.is_empty() {
+            let status = std::process::Command::new("netsh")
+                .args(["interface", "ipv4", "set", "dnsservers", name, "dhcp"])
+                .status()
+                .map_err(|e| format!("failed to restore DNS for {}: {}", name, e))?;
+
+            if status.success() {
+                eprintln!("  restored DNS for \"{}\" -> DHCP", name);
+            } else {
+                eprintln!("  warning: failed to restore DNS for \"{}\"", name);
+            }
+        } else {
+            let status = std::process::Command::new("netsh")
+                .args([
+                    "interface",
+                    "ipv4",
+                    "set",
+                    "dnsservers",
+                    name,
+                    "static",
+                    &dns_info.servers[0],
+                    "primary",
+                ])
+                .status()
+                .map_err(|e| format!("failed to restore DNS for {}: {}", name, e))?;
+
+            if !status.success() {
+                eprintln!("  warning: failed to restore primary DNS for \"{}\"", name);
+                continue;
+            }
+
+            for (i, server) in dns_info.servers.iter().skip(1).enumerate() {
+                let _ = std::process::Command::new("netsh")
+                    .args([
+                        "interface",
+                        "ipv4",
+                        "add",
+                        "dnsservers",
+                        name,
+                        server,
+                        &format!("index={}", i + 2),
+                    ])
+                    .status();
+            }
+
+            eprintln!(
+                "  restored DNS for \"{}\" -> {}",
+                name,
+                dns_info.servers.join(", ")
+            );
+        }
+    }
+
+    std::fs::remove_file(&path).ok();
+
+    // Re-enable Dnscache
+    enable_dnscache();
+    eprintln!("\n  System DNS restored. DNS Client re-enabled.");
+    eprintln!("  Reboot to fully restore the DNS Client service.\n");
+    Ok(())
+}
+
 /// Find the upstream for a domain by checking forwarding rules.
 /// Returns None if no rule matches (use default upstream).
 /// Zero-allocation on the hot path — dot_suffix is pre-computed.
@@ -316,43 +705,6 @@ pub fn match_forwarding_rule(domain: &str, rules: &[ForwardingRule]) -> Option<S
 
 // --- System DNS configuration (install/uninstall) ---
 
-/// Set the system DNS to 127.0.0.1 so all queries go through Numa.
-/// Saves the original DNS settings for later restoration.
-pub fn install_system_dns() -> Result<(), String> {
-    #[cfg(target_os = "macos")]
-    let result = install_macos();
-    #[cfg(target_os = "linux")]
-    let result = install_linux();
-    #[cfg(not(any(target_os = "macos", target_os = "linux")))]
-    let result = Err("system DNS configuration not supported on this OS".to_string());
-
-    if result.is_ok() {
-        if let Err(e) = trust_ca() {
-            eprintln!("  warning: could not trust CA: {}", e);
-            eprintln!("  HTTPS proxy will work but browsers will show certificate warnings.\n");
-        }
-    }
-    result
-}
-
-/// Restore the original system DNS settings saved during install.
-pub fn uninstall_system_dns() -> Result<(), String> {
-    let _ = untrust_ca();
-
-    #[cfg(target_os = "macos")]
-    {
-        uninstall_macos()
-    }
-    #[cfg(target_os = "linux")]
-    {
-        uninstall_linux()
-    }
-    #[cfg(not(any(target_os = "macos", target_os = "linux")))]
-    {
-        Err("system DNS configuration not supported on this OS".to_string())
-    }
-}
-
 // --- macOS implementation ---
 
 #[cfg(target_os = "macos")]
@@ -424,7 +776,7 @@ fn install_macos() -> Result<(), String> {
         .map_err(|e| format!("failed to serialize backup: {}", e))?;
     std::fs::write(backup_path(), json).map_err(|e| format!("failed to write backup: {}", e))?;
 
-    // Set DNS to 127.0.0.1 for each service
+    // Set DNS to 127.0.0.1 and add "numa" search domain for each service
     for service in &services {
         let status = std::process::Command::new("networksetup")
             .args(["-setdnsservers", service, "127.0.0.1"])
@@ -436,6 +788,11 @@ fn install_macos() -> Result<(), String> {
         } else {
             eprintln!("  warning: failed to set DNS for \"{}\"", service);
         }
+
+        // Add "numa" as search domain so browsers resolve .numa without trailing slash
+        let _ = std::process::Command::new("networksetup")
+            .args(["-setsearchdomains", service, "numa"])
+            .status();
     }
 
     eprintln!("\n  Original DNS saved to {}", backup_path().display());
@@ -480,6 +837,11 @@ fn uninstall_macos() -> Result<(), String> {
         } else {
             eprintln!("  warning: failed to restore DNS for \"{}\"", service);
         }
+
+        // Clear the "numa" search domain
+        let _ = std::process::Command::new("networksetup")
+            .args(["-setsearchdomains", service, "Empty"])
+            .status();
     }
 
     std::fs::remove_file(&path).ok();
@@ -500,21 +862,27 @@ const SYSTEMD_UNIT: &str = "/etc/systemd/system/numa.service";
 /// Install Numa as a system service that starts on boot and auto-restarts.
 pub fn install_service() -> Result<(), String> {
     #[cfg(target_os = "macos")]
-    {
-        install_service_macos()
-    }
+    let result = install_service_macos();
     #[cfg(target_os = "linux")]
-    {
-        install_service_linux()
-    }
-    #[cfg(not(any(target_os = "macos", target_os = "linux")))]
-    {
-        Err("service installation not supported on this OS".to_string())
+    let result = install_service_linux();
+    #[cfg(windows)]
+    let result = install_windows();
+    #[cfg(not(any(target_os = "macos", target_os = "linux", windows)))]
+    let result = Err::<(), String>("service installation not supported on this OS".to_string());
+
+    if result.is_ok() {
+        if let Err(e) = trust_ca() {
+            eprintln!("  warning: could not trust CA: {}", e);
+            eprintln!("  HTTPS proxy will work but browsers will show certificate warnings.\n");
+        }
     }
+    result
 }
 
 /// Uninstall the Numa system service.
 pub fn uninstall_service() -> Result<(), String> {
+    let _ = untrust_ca();
+
     #[cfg(target_os = "macos")]
     {
         uninstall_service_macos()
@@ -523,7 +891,11 @@ pub fn uninstall_service() -> Result<(), String> {
     {
         uninstall_service_linux()
     }
-    #[cfg(not(any(target_os = "macos", target_os = "linux")))]
+    #[cfg(windows)]
+    {
+        uninstall_windows()
+    }
+    #[cfg(not(any(target_os = "macos", target_os = "linux", windows)))]
     {
         Err("service uninstallation not supported on this OS".to_string())
     }
@@ -531,9 +903,13 @@ pub fn uninstall_service() -> Result<(), String> {
 
 /// Restart the service (kill process, launchd/systemd auto-restarts with new binary).
 pub fn restart_service() -> Result<(), String> {
+    #[cfg(any(target_os = "macos", target_os = "linux"))]
+    let exe_path =
+        std::env::current_exe().map_err(|e| format!("failed to get current exe: {}", e))?;
+
     #[cfg(any(target_os = "macos", target_os = "linux"))]
     let version = {
-        match std::process::Command::new("/usr/local/bin/numa")
+        match std::process::Command::new(&exe_path)
             .arg("--version")
             .output()
         {
@@ -544,6 +920,7 @@ pub fn restart_service() -> Result<(), String> {
 
     #[cfg(target_os = "macos")]
     {
+        let exe_path = exe_path.to_string_lossy();
         let output = std::process::Command::new("launchctl")
             .args(["list", PLIST_LABEL])
             .output();
@@ -554,11 +931,11 @@ pub fn restart_service() -> Result<(), String> {
                 // This will kill us too (we ARE /usr/local/bin/numa), so
                 // codesign and print output first.
                 let _ = std::process::Command::new("codesign")
-                    .args(["-f", "-s", "-", "/usr/local/bin/numa"])
+                    .args(["-f", "-s", "-", &exe_path])
                     .output(); // use output() to suppress codesign stderr
                 eprintln!("  Service restarting → {}\n", version);
                 let _ = std::process::Command::new("pkill")
-                    .args(["-f", "/usr/local/bin/numa"])
+                    .args(["-f", &exe_path])
                     .status();
                 Ok(())
             }
@@ -593,23 +970,27 @@ pub fn service_status() -> Result<(), String> {
     }
 }
 
+#[cfg(any(target_os = "macos", target_os = "linux"))]
+fn replace_exe_path(service: &str) -> Result<String, String> {
+    let exe_path =
+        std::env::current_exe().map_err(|e| format!("failed to get current exe: {}", e))?;
+    Ok(service.replace("{{exe_path}}", &exe_path.to_string_lossy()))
+}
+
 #[cfg(target_os = "macos")]
 fn install_service_macos() -> Result<(), String> {
-    // Check binary exists
-    if !std::path::Path::new("/usr/local/bin/numa").exists() {
-        return Err("numa binary not found at /usr/local/bin/numa. Run: sudo cp target/release/numa /usr/local/bin/numa".to_string());
-    }
-
     // Create log directory
     std::fs::create_dir_all("/usr/local/var/log")
         .map_err(|e| format!("failed to create log dir: {}", e))?;
 
     // Write plist
     let plist = include_str!("../com.numa.dns.plist");
+    let plist = replace_exe_path(plist)?;
+
     std::fs::write(PLIST_DEST, plist)
         .map_err(|e| format!("failed to write {}: {}", PLIST_DEST, e))?;
 
-    // Load the service
+    // Load the service first so numa is listening before DNS redirect
     let status = std::process::Command::new("launchctl")
         .args(["load", "-w", PLIST_DEST])
         .status()
@@ -619,14 +1000,34 @@ fn install_service_macos() -> Result<(), String> {
         return Err("launchctl load failed".to_string());
     }
 
-    // Set system DNS to 127.0.0.1 now that the service is running
-    eprintln!("  Service installed and started.");
+    // Wait for numa to be ready before redirecting DNS
+    let api_up = (0..10).any(|i| {
+        if i > 0 {
+            std::thread::sleep(std::time::Duration::from_millis(500));
+        }
+        std::net::TcpStream::connect(("127.0.0.1", crate::config::DEFAULT_API_PORT)).is_ok()
+    });
+    if !api_up {
+        // Service failed to start — don't redirect DNS to a dead endpoint
+        let _ = std::process::Command::new("launchctl")
+            .args(["unload", PLIST_DEST])
+            .status();
+        return Err(
+            "numa service did not start (port 53 may be in use). Service unloaded.".to_string(),
+        );
+    }
+
     if let Err(e) = install_macos() {
         eprintln!("  warning: failed to configure system DNS: {}", e);
     }
+
+    eprintln!("  Service installed and started.");
     eprintln!("  Numa will auto-start on boot and restart if killed.");
     eprintln!("  Logs: /usr/local/var/log/numa.log");
-    eprintln!("  Run 'sudo numa service stop' to fully uninstall.\n");
+    eprintln!("  Run 'sudo numa uninstall' to restore original DNS.\n");
+    eprintln!("  Want full DNS sovereignty? Add to numa.toml:");
+    eprintln!("    [upstream]");
+    eprintln!("    mode = \"recursive\"\n");
     Ok(())
 }
 
@@ -708,8 +1109,11 @@ fn install_linux() -> Result<(), String> {
             .map_err(|e| format!("failed to create {}: {}", resolved_dir.display(), e))?;
 
         let drop_in = resolved_dir.join("numa.conf");
-        std::fs::write(&drop_in, "[Resolve]\nDNS=127.0.0.1\nDomains=~.\n")
-            .map_err(|e| format!("failed to write {}: {}", drop_in.display(), e))?;
+        std::fs::write(
+            &drop_in,
+            "[Resolve]\nDNS=127.0.0.1\nDomains=~. numa\nDNSStubListener=no\n",
+        )
+        .map_err(|e| format!("failed to write {}: {}", drop_in.display(), e))?;
 
         let _ = run_systemctl(&["restart", "systemd-resolved"]);
         eprintln!("  systemd-resolved detected.");
@@ -745,7 +1149,7 @@ fn install_linux() -> Result<(), String> {
     }
 
     let content =
-        "# Generated by Numa — run 'sudo numa uninstall' to restore\nnameserver 127.0.0.1\n";
+        "# Generated by Numa — run 'sudo numa uninstall' to restore\nnameserver 127.0.0.1\nsearch numa\n";
     std::fs::write(resolv, content)
         .map_err(|e| format!("failed to write /etc/resolv.conf: {}", e))?;
 
@@ -784,35 +1188,30 @@ fn uninstall_linux() -> Result<(), String> {
     Ok(())
 }
 
-#[cfg(target_os = "linux")]
-fn ensure_binary_installed() -> Result<(), String> {
-    if !std::path::Path::new("/usr/local/bin/numa").exists() {
-        return Err("numa binary not found at /usr/local/bin/numa. Run: sudo cp target/release/numa /usr/local/bin/numa".to_string());
-    }
-    Ok(())
-}
-
 #[cfg(target_os = "linux")]
 fn install_service_linux() -> Result<(), String> {
-    ensure_binary_installed()?;
-
     let unit = include_str!("../numa.service");
+    let unit = replace_exe_path(unit)?;
     std::fs::write(SYSTEMD_UNIT, unit)
         .map_err(|e| format!("failed to write {}: {}", SYSTEMD_UNIT, e))?;
 
     run_systemctl(&["daemon-reload"])?;
     run_systemctl(&["enable", "numa"])?;
-    run_systemctl(&["start", "numa"])?;
 
-    eprintln!("  Service installed and started.");
-
-    // Set system DNS now that the service is running
+    // Configure system DNS before starting numa so resolved releases port 53 first
     if let Err(e) = install_linux() {
         eprintln!("  warning: failed to configure system DNS: {}", e);
     }
+
+    run_systemctl(&["start", "numa"])?;
+
+    eprintln!("  Service installed and started.");
     eprintln!("  Numa will auto-start on boot and restart if killed.");
     eprintln!("  Logs: journalctl -u numa -f");
-    eprintln!("  Run 'sudo numa service stop' to fully uninstall.\n");
+    eprintln!("  Run 'sudo numa uninstall' to restore original DNS.\n");
+    eprintln!("  Want full DNS sovereignty? Add to numa.toml:");
+    eprintln!("    [upstream]");
+    eprintln!("    mode = \"recursive\"\n");
     Ok(())
 }
 
@@ -977,3 +1376,76 @@ fn untrust_ca() -> Result<(), String> {
     let _ = ca_path; // suppress unused warning on other platforms
     Ok(())
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn parse_ipconfig_dhcp_and_static() {
+        let sample = "\
+Ethernet adapter Ethernet:
+
+   DHCP Enabled. . . . . . . . . . . : Yes
+   DNS Servers . . . . . . . . . . . : 8.8.8.8
+                                        8.8.4.4
+
+Wireless LAN adapter Wi-Fi:
+
+   DHCP Enabled. . . . . . . . . . . : No
+   DNS Servers . . . . . . . . . . . : 1.1.1.1
+";
+        let result = parse_ipconfig_interfaces(sample);
+        assert_eq!(result.len(), 2);
+        assert_eq!(
+            result["Ethernet"],
+            WindowsInterfaceDns {
+                dhcp: true,
+                servers: vec!["8.8.8.8".into(), "8.8.4.4".into()],
+            }
+        );
+        assert_eq!(
+            result["Wi-Fi"],
+            WindowsInterfaceDns {
+                dhcp: false,
+                servers: vec!["1.1.1.1".into()],
+            }
+        );
+    }
+
+    #[test]
+    #[cfg(any(target_os = "macos", target_os = "linux"))]
+    fn replace_exe_path_substitutes_template() {
+        let plist = include_str!("../com.numa.dns.plist");
+        let unit = include_str!("../numa.service");
+
+        assert!(plist.contains("{{exe_path}}"), "plist missing placeholder");
+        assert!(
+            unit.contains("{{exe_path}}"),
+            "unit file missing placeholder"
+        );
+
+        let result = replace_exe_path(plist).expect("replace_exe_path failed for plist");
+        assert!(!result.contains("{{exe_path}}"));
+
+        let result = replace_exe_path(unit).expect("replace_exe_path failed for unit");
+        assert!(!result.contains("{{exe_path}}"));
+    }
+
+    #[test]
+    fn parse_ipconfig_skips_disconnected() {
+        let sample = "\
+Ethernet adapter Ethernet 2:
+
+   Media State . . . . . . . . . . . : Media disconnected
+
+Wireless LAN adapter Wi-Fi:
+
+   DHCP Enabled. . . . . . . . . . . : Yes
+   DNS Servers . . . . . . . . . . . : 192.168.1.1
+";
+        let result = parse_ipconfig_interfaces(sample);
+        assert_eq!(result.len(), 1);
+        assert!(result.contains_key("Wi-Fi"));
+    }
+}

src/tls.rs

@@ -24,7 +24,7 @@ pub fn regenerate_tls(ctx: &ServerCtx) {
     names.extend(ctx.lan_peers.lock().unwrap().names());
     let names: Vec<String> = names.into_iter().collect();
 
-    match build_tls_config(&ctx.proxy_tld, &names) {
+    match build_tls_config(&ctx.proxy_tld, &names, Vec::new(), &ctx.data_dir) {
         Ok(new_config) => {
             tls.store(new_config);
             info!("TLS cert regenerated for {} services", names.len());
@@ -36,17 +36,26 @@ pub fn regenerate_tls(ctx: &ServerCtx) {
 /// Build a TLS config with a cert covering all provided service names.
 /// Wildcards under single-label TLDs (*.numa) are rejected by browsers,
 /// so we list each service explicitly as a SAN.
-pub fn build_tls_config(tld: &str, service_names: &[String]) -> crate::Result<Arc<ServerConfig>> {
-    let dir = crate::data_dir();
-    let (ca_cert, ca_key) = ensure_ca(&dir)?;
+/// `alpn` is advertised in the TLS ServerHello — pass empty for the proxy
+/// (which accepts any ALPN), or `[b"dot"]` for DoT (RFC 7858 §3.2).
+/// `data_dir` is where the CA material is stored — taken from
+/// `[server] data_dir` in numa.toml (defaults to `crate::data_dir()`).
+pub fn build_tls_config(
+    tld: &str,
+    service_names: &[String],
+    alpn: Vec<Vec<u8>>,
+    data_dir: &Path,
+) -> crate::Result<Arc<ServerConfig>> {
+    let (ca_cert, ca_key) = ensure_ca(data_dir)?;
     let (cert_chain, key) = generate_service_cert(&ca_cert, &ca_key, tld, service_names)?;
 
     // Ensure a crypto provider is installed (rustls needs one)
     let _ = rustls::crypto::ring::default_provider().install_default();
 
-    let config = ServerConfig::builder()
+    let mut config = ServerConfig::builder()
         .with_no_client_auth()
         .with_single_cert(cert_chain, key)?;
+    config.alpn_protocols = alpn;
 
     info!(
         "TLS configured for {} .{} domains",

tests/integration.sh

@@ -404,6 +404,241 @@ check "Cache flushed" \
 
 kill "$NUMA_PID" 2>/dev/null || true
 wait "$NUMA_PID" 2>/dev/null || true
+sleep 1
+
+# ---- Suite 5: DNS-over-TLS (RFC 7858) ----
+echo ""
+echo "╔══════════════════════════════════════════╗"
+echo "║  Suite 5: DNS-over-TLS (RFC 7858)        ║"
+echo "╚══════════════════════════════════════════╝"
+
+if ! command -v kdig >/dev/null 2>&1; then
+    printf "  ${DIM}skipped — install 'knot' for kdig${RESET}\n"
+elif ! command -v openssl >/dev/null 2>&1; then
+    printf "  ${DIM}skipped — openssl not found${RESET}\n"
+else
+    DOT_PORT=8853
+    DOT_CERT=/tmp/numa-integration-dot.crt
+    DOT_KEY=/tmp/numa-integration-dot.key
+
+    # Generate a test cert mirroring production self_signed_tls SAN shape
+    # (*.numa wildcard + explicit numa.numa apex).
+    openssl req -x509 -newkey rsa:2048 -nodes -days 1 \
+        -keyout "$DOT_KEY" -out "$DOT_CERT" \
+        -subj "/CN=Numa .numa services" \
+        -addext "subjectAltName=DNS:*.numa,DNS:numa.numa" \
+        >/dev/null 2>&1
+
+    # Suite 5 uses a local zone so it's upstream-independent — the point is
+    # to exercise the DoT transport layer (handshake, ALPN, framing,
+    # persistent connections), not re-test recursive resolution.
+    cat > "$CONFIG" << CONF
+[server]
+bind_addr = "127.0.0.1:$PORT"
+api_port = $API_PORT
+
+[upstream]
+mode = "forward"
+address = "127.0.0.1"
+port = 65535
+
+[cache]
+max_entries = 10000
+
+[blocking]
+enabled = false
+
+[proxy]
+enabled = false
+
+[dot]
+enabled = true
+port = $DOT_PORT
+bind_addr = "127.0.0.1"
+cert_path = "$DOT_CERT"
+key_path = "$DOT_KEY"
+
+[[zones]]
+domain = "dot-test.example"
+record_type = "A"
+value = "10.0.0.1"
+ttl = 60
+CONF
+
+    RUST_LOG=info "$BINARY" "$CONFIG" > "$LOG" 2>&1 &
+    NUMA_PID=$!
+    sleep 4
+
+    if ! kill -0 "$NUMA_PID" 2>/dev/null; then
+        FAILED=$((FAILED + 1))
+        printf "  ${RED}✗${RESET} DoT startup\n"
+        printf "    ${DIM}%s${RESET}\n" "$(tail -5 "$LOG")"
+    else
+        echo ""
+        echo "=== Listener ==="
+
+        check "DoT bound on 127.0.0.1:$DOT_PORT" \
+            "DoT listening on 127.0.0.1:$DOT_PORT" \
+            "$(grep 'DoT listening' "$LOG")"
+
+        KDIG="kdig @127.0.0.1 -p $DOT_PORT +tls +tls-ca=$DOT_CERT +tls-hostname=numa.numa +time=5 +retry=0"
+
+        echo ""
+        echo "=== Queries over DoT ==="
+
+        check "DoT local zone A record" \
+            "10.0.0.1" \
+            "$($KDIG +short dot-test.example A 2>/dev/null)"
+
+        # +keepopen reuses one TLS connection for multiple queries — tests
+        # persistent connection handling. kdig applies options left-to-right,
+        # so +short and +keepopen must come before the query specs.
+        check "DoT persistent connection (3 queries, 1 handshake)" \
+            "10.0.0.1" \
+            "$($KDIG +keepopen +short dot-test.example A dot-test.example A dot-test.example A 2>/dev/null | head -1)"
+
+        echo ""
+        echo "=== ALPN ==="
+
+        # Positive case: client offers "dot", server picks it.
+        ALPN_OK=$(echo "" | openssl s_client -connect "127.0.0.1:$DOT_PORT" \
+            -servername numa.numa -alpn dot -CAfile "$DOT_CERT" 2>&1 </dev/null || true)
+        check "DoT negotiates ALPN \"dot\"" \
+            "ALPN protocol: dot" \
+            "$ALPN_OK"
+
+        # Negative case: client offers only "h2", server must reject the
+        # handshake with no_application_protocol alert (cross-protocol
+        # confusion defense, RFC 7858bis §3.2).
+        if echo "" | openssl s_client -connect "127.0.0.1:$DOT_PORT" \
+            -servername numa.numa -alpn h2 -CAfile "$DOT_CERT" \
+            </dev/null >/dev/null 2>&1; then
+            ALPN_MISMATCH="handshake unexpectedly succeeded"
+        else
+            ALPN_MISMATCH="rejected"
+        fi
+        check "DoT rejects non-dot ALPN" \
+            "rejected" \
+            "$ALPN_MISMATCH"
+    fi
+
+    kill "$NUMA_PID" 2>/dev/null || true
+    wait "$NUMA_PID" 2>/dev/null || true
+    rm -f "$DOT_CERT" "$DOT_KEY"
+fi
+sleep 1
+
+# ---- Suite 6: Proxy + DoT coexistence ----
+echo ""
+echo "╔══════════════════════════════════════════╗"
+echo "║  Suite 6: Proxy + DoT Coexistence        ║"
+echo "╚══════════════════════════════════════════╝"
+
+if ! command -v kdig >/dev/null 2>&1 || ! command -v openssl >/dev/null 2>&1; then
+    printf "  ${DIM}skipped — needs kdig + openssl${RESET}\n"
+else
+    DOT_PORT=8853
+    PROXY_HTTP_PORT=8080
+    PROXY_HTTPS_PORT=8443
+    NUMA_DATA=/tmp/numa-integration-data
+
+    # Fresh data dir so we generate a fresh CA for this suite. Path is set
+    # via [server] data_dir in the TOML below, not an env var — numa treats
+    # its config file as the single source of truth for all knobs.
+    rm -rf "$NUMA_DATA"
+    mkdir -p "$NUMA_DATA"
+
+    cat > "$CONFIG" << CONF
+[server]
+bind_addr = "127.0.0.1:$PORT"
+api_port = $API_PORT
+data_dir = "$NUMA_DATA"
+
+[upstream]
+mode = "forward"
+address = "127.0.0.1"
+port = 65535
+
+[cache]
+max_entries = 10000
+
+[blocking]
+enabled = false
+
+[proxy]
+enabled = true
+port = $PROXY_HTTP_PORT
+tls_port = $PROXY_HTTPS_PORT
+tld = "numa"
+bind_addr = "127.0.0.1"
+
+[dot]
+enabled = true
+port = $DOT_PORT
+bind_addr = "127.0.0.1"
+
+[[zones]]
+domain = "dot-test.example"
+record_type = "A"
+value = "10.0.0.1"
+ttl = 60
+CONF
+
+    RUST_LOG=info "$BINARY" "$CONFIG" > "$LOG" 2>&1 &
+    NUMA_PID=$!
+    sleep 4
+
+    if ! kill -0 "$NUMA_PID" 2>/dev/null; then
+        FAILED=$((FAILED + 1))
+        printf "  ${RED}✗${RESET} Startup with proxy + DoT\n"
+        printf "    ${DIM}%s${RESET}\n" "$(tail -5 "$LOG")"
+    else
+        echo ""
+        echo "=== Both listeners ==="
+
+        check "DoT listener bound" \
+            "DoT listening on 127.0.0.1:$DOT_PORT" \
+            "$(grep 'DoT listening' "$LOG")"
+
+        check "HTTPS proxy listener bound" \
+            "HTTPS proxy listening on 127.0.0.1:$PROXY_HTTPS_PORT" \
+            "$(grep 'HTTPS proxy listening' "$LOG")"
+
+        PANIC_COUNT=$(grep -c 'panicked' "$LOG" 2>/dev/null || echo 0)
+        check "No startup panics in log" \
+            "^0$" \
+            "$PANIC_COUNT"
+
+        echo ""
+        echo "=== DoT works with proxy enabled ==="
+
+        # Proxy's build_tls_config runs first and creates the CA in
+        # $NUMA_DATA_DIR. DoT self_signed_tls then loads the same CA and
+        # issues its own leaf cert. One CA trusts both listeners.
+        CA="$NUMA_DATA/ca.pem"
+        KDIG="kdig @127.0.0.1 -p $DOT_PORT +tls +tls-ca=$CA +tls-hostname=numa.numa +time=5 +retry=0"
+
+        check "DoT local zone A (with proxy on)" \
+            "10.0.0.1" \
+            "$($KDIG +short dot-test.example A 2>/dev/null)"
+
+        echo ""
+        echo "=== Proxy TLS works with DoT enabled ==="
+
+        # Proxy cert has SAN numa.numa (auto-added "numa" service). A
+        # successful handshake validates that the proxy's separate
+        # ServerConfig wasn't disturbed by DoT's own cert generation.
+        PROXY_TLS=$(echo "" | openssl s_client -connect "127.0.0.1:$PROXY_HTTPS_PORT" \
+            -servername numa.numa -CAfile "$CA" 2>&1 </dev/null || true)
+        check "Proxy HTTPS TLS handshake succeeds" \
+            "Verify return code: 0 (ok)" \
+            "$PROXY_TLS"
+    fi
+
+    kill "$NUMA_PID" 2>/dev/null || true
+    wait "$NUMA_PID" 2>/dev/null || true
+    rm -rf "$NUMA_DATA"
+fi
 
 # Summary
 echo ""