chore: bump version to 0.13.1

fix: accept tls:// and https:// in [[forwarding]] upstreams

.github/workflows/ci.yml

@@ -3,8 +3,22 @@ name: CI
 on:
   push:
     branches: [main]
+    paths-ignore:
+      - 'site/**'
+      - 'blog/**'
+      - 'drafts/**'
+      - '*.md'
+      - 'scripts/serve-site.sh'
+      - 'scripts/generate-blog-index.sh'
   pull_request:
     branches: [main]
+    paths-ignore:
+      - 'site/**'
+      - 'blog/**'
+      - 'drafts/**'
+      - '*.md'
+      - 'scripts/serve-site.sh'
+      - 'scripts/generate-blog-index.sh'
 
 env:
   CARGO_TERM_COLOR: always

.github/workflows/docker.yml

@@ -0,0 +1,45 @@
+name: Docker
+
+on:
+  push:
+    tags:
+      - 'v*'
+
+permissions:
+  contents: read
+  packages: write
+
+jobs:
+  docker:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v6
+
+      - uses: docker/setup-qemu-action@v3
+
+      - uses: docker/setup-buildx-action@v3
+
+      - uses: docker/login-action@v3
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - uses: docker/metadata-action@v5
+        id: meta
+        with:
+          images: ghcr.io/${{ github.repository }}
+          tags: |
+            type=semver,pattern={{version}}
+            type=semver,pattern={{major}}.{{minor}}
+            type=raw,value=latest
+
+      - uses: docker/build-push-action@v6
+        with:
+          context: .
+          platforms: linux/amd64,linux/arm64
+          push: true
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max

.github/workflows/publish-aur.yml

@@ -23,6 +23,13 @@ name: Publish - Arch Linux AUR Package
 on:
   push:
     branches: [main]
+    paths-ignore:
+      - 'site/**'
+      - 'blog/**'
+      - 'drafts/**'
+      - '*.md'
+      - 'scripts/serve-site.sh'
+      - 'scripts/generate-blog-index.sh'
   workflow_dispatch:
 
 permissions:

.gitignore

@@ -4,3 +4,5 @@ CLAUDE.md
 docs/
 site/blog/posts/
 ios/
+drafts/
+site/blog/index.html

Cargo.lock

@@ -1330,7 +1330,7 @@ dependencies = [
 
 [[package]]
 name = "numa"
-version = "0.13.0"
+version = "0.13.1"
 dependencies = [
  "arc-swap",
  "axum",

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "numa"
-version = "0.13.0"
+version = "0.13.1"
 authors = ["razvandimescu <razvan@dimescu.com>"]
 edition = "2021"
 description = "Portable DNS resolver in Rust — .numa local domains, ad blocking, developer overrides, DNS-over-HTTPS"

Dockerfile

@@ -6,6 +6,7 @@ RUN mkdir src && echo 'fn main() {}' > src/main.rs && echo '' > src/lib.rs
 RUN cargo build --release 2>/dev/null || true
 RUN rm -rf src
 COPY src/ src/
+COPY benches/ benches/
 COPY site/ site/
 COPY numa.toml com.numa.dns.plist numa.service ./
 RUN touch src/main.rs src/lib.rs
@@ -13,5 +14,6 @@ RUN cargo build --release
 
 FROM alpine:3.23
 COPY --from=builder /app/target/release/numa /usr/local/bin/numa
+RUN mkdir -p /root/.config/numa && printf '[server]\napi_bind_addr = "0.0.0.0"\n\n[proxy]\nenabled = true\nbind_addr = "0.0.0.0"\n' > /root/.config/numa/numa.toml
 EXPOSE 53/udp 80/tcp 443/tcp 853/tcp 5380/tcp
 ENTRYPOINT ["numa"]

Makefile

@@ -32,6 +32,19 @@ blog:
 		pandoc "$$f" --template=site/blog-template.html -o "site/blog/posts/$$name.html"; \
 		echo "  $$f → site/blog/posts/$$name.html"; \
 	done
+	@scripts/generate-blog-index.sh
+
+blog-drafts: blog
+	@if [ -d drafts ] && ls drafts/*.md >/dev/null 2>&1; then \
+		for f in drafts/*.md; do \
+			name=$$(basename "$$f" .md); \
+			pandoc "$$f" --template=site/blog-template.html -o "site/blog/posts/$$name.html"; \
+			echo "  $$f → site/blog/posts/$$name.html (draft)"; \
+		done; \
+		BLOG_INCLUDE_DRAFTS=1 scripts/generate-blog-index.sh; \
+	else \
+		echo "  No drafts found"; \
+	fi
 
 release:
 ifndef VERSION

PKGBUILD

@@ -9,7 +9,7 @@ url="https://github.com/razvandimescu/numa"
 license=('MIT')
 options=('!lto')
 depends=('gcc-libs' 'glibc')
-makedepends=('cargo' 'git')
+makedepends=('cargo' 'git' 'llvm-libs')
 provides=("$_pkgname")
 conflicts=("$_pkgname")
 backup=('etc/numa.toml')

README.md

@@ -27,6 +27,9 @@ yay -S numa-git
 # Windows — download from GitHub Releases
 # All platforms
 cargo install numa
+
+# Docker
+docker run -d --name numa --network host ghcr.io/razvandimescu/numa
 ```
 
 ```bash
@@ -102,6 +105,26 @@ From Machine B: `curl http://api.numa` → proxied to Machine A's port 8000. Ena
 
 **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.
 
+## Docker
+
+```bash
+# Recommended — host networking (Linux)
+docker run -d --name numa --network host ghcr.io/razvandimescu/numa
+
+# Port mapping (macOS/Windows Docker Desktop)
+docker run -d --name numa -p 53:53/udp -p 53:53/tcp -p 5380:5380 ghcr.io/razvandimescu/numa
+```
+
+Dashboard at `http://localhost:5380`. The image binds the API and proxy to `0.0.0.0` by default. Override with a custom config:
+
+```bash
+docker run -d --name numa --network host \
+  -v /path/to/numa.toml:/root/.config/numa/numa.toml \
+  ghcr.io/razvandimescu/numa
+```
+
+Multi-arch: `linux/amd64` and `linux/arm64`.
+
 ## How It Compares
 
 | | Pi-hole | AdGuard Home | Unbound | Numa |

blog/dns-from-scratch.md

@@ -1,7 +1,7 @@
 ---
 title: I Built a DNS Resolver from Scratch in Rust
 description: How DNS actually works at the wire level — label compression, TTL tricks, DoH, and what surprised me building a resolver with zero DNS libraries.
-date: March 2026
+date: 2026-03-20
 ---
 
 I wanted to understand how DNS actually works. Not the "it translates domain names to IP addresses" explanation — the actual bytes on the wire. What does a DNS packet look like? How does label compression work? Why is everything crammed into 512 bytes?

blog/dnssec-from-scratch.md

@@ -1,7 +1,7 @@
 ---
 title: Implementing DNSSEC from Scratch in Rust
 description: Recursive resolution from root hints, chain-of-trust validation, NSEC/NSEC3 denial proofs, and what I learned implementing DNSSEC with zero DNS libraries.
-date: March 2026
+date: 2026-03-28
 ---
 
 In the [previous post](/blog/posts/dns-from-scratch.html) I covered how DNS works at the wire level — packet format, label compression, TTL caching, DoH. Numa was a forwarding resolver: it parsed packets, did useful things locally, and relayed the rest to Cloudflare or Quad9.

blog/dot-from-scratch.md

@@ -1,7 +1,7 @@
 ---
 title: DNS-over-TLS from Scratch in Rust
 description: Building RFC 7858 on top of rustls — length-prefix framing, ALPN cross-protocol defense, and two bugs that only the strict clients caught.
-date: April 2026
+date: 2026-04-06
 ---
 
 The [previous post](/blog/posts/dnssec-from-scratch.html) ended with "DoT — the last encrypted transport we don't support." This post is about building it.

blog/fixing-doh-tail-latency.md

@@ -0,0 +1,171 @@
+---
+title: Fixing DNS tail latency with a 5-line config and a 50-line function
+description: Periodic 40-140ms DoH spikes from hyper's dispatch channel. The fix was reqwest window tuning and request hedging — Dean & Barroso's "The Tail at Scale," applied to a DNS forwarder. Same ideas took cold recursive p99 from 2.3 seconds to 538ms.
+date: 2026-04-12
+---
+
+If you're using reqwest for small HTTP/2 payloads, you probably have a tail latency problem you don't know about. Hyper's default flow control windows are 10,000× oversized for anything under 1 KB, and its dispatch channel adds periodic 40-140ms stalls that don't show up in median benchmarks.
+
+I hit this building Numa's DoH forwarding path. Median was 10ms, mean was 23ms — the tail was dragging everything.
+
+<div class="hero-metrics">
+<div class="metric-card">
+<div class="metric-vs">DoH forwarding p99</div>
+<div class="metric-value">113 → 71ms</div>
+<div class="metric-label">window tuning + request hedging</div>
+</div>
+<div class="metric-card">
+<div class="metric-vs">Cold recursive p99</div>
+<div class="metric-value">2.3s → 538ms</div>
+<div class="metric-label">NS caching, serve-stale, parallel queries</div>
+</div>
+<div class="metric-card">
+<div class="metric-vs">Forwarding σ</div>
+<div class="metric-value">31 → 13ms</div>
+<div class="metric-label">random spikes become parallel races</div>
+</div>
+</div>
+
+The fix was a 5-line reqwest config and a 50-line hedging function. This post is also an advertisement for Dean & Barroso's 2013 paper ["The Tail at Scale"](https://research.google/pubs/pub40801/) — a decade-old idea that still demolishes dispatch spikes. The same ideas later took my cold recursive p99 from 2.3 seconds to 538ms.
+
+---
+
+## The cause: hyper's dispatch channel
+
+Reqwest sits on top of hyper, which interposes an mpsc dispatch channel and a separate `ClientTask` between `.send()` and the h2 stream. I instrumented the forwarding path and confirmed: 100% of the spike time lives in the `send()` phase, and a parallel heartbeat task showed zero runtime lag during spikes. The tokio runtime was fine — the stall was internal to hyper's request scheduling.
+
+Hickory-resolver doesn't have this issue. It holds `h2::SendRequest<Bytes>` directly and calls `ready().await; send_request()` in the caller's task — no channel, no scheduling dependency. I used it as a reference point throughout.
+
+## Fix #1 — HTTP/2 window sizes
+
+Reqwest inherits hyper's HTTP/2 defaults: 2 MB stream window, 5 MB connection window. For DNS responses (~200 bytes), that's ~10,000× oversized — unnecessary WINDOW_UPDATE frames, bloated bookkeeping on every poll, and different server-side scheduling behavior.
+
+Setting both windows to the h2 spec default (64 KB) dropped my median from 13.3ms to 10.1ms:
+
+```rust
+reqwest::Client::builder()
+    .use_rustls_tls()
+    .http2_initial_stream_window_size(65_535)
+    .http2_initial_connection_window_size(65_535)
+    .http2_keep_alive_interval(Duration::from_secs(15))
+    .http2_keep_alive_while_idle(true)
+    .http2_keep_alive_timeout(Duration::from_secs(10))
+    .pool_idle_timeout(Duration::from_secs(300))
+    .pool_max_idle_per_host(1)
+    .build()
+```
+
+**Any Rust code using reqwest for tiny-payload HTTP/2 workloads — DoH, API polling, metric scraping — is probably hitting this.**
+
+## Fix #2 — Request hedging
+
+["The Tail at Scale"](https://research.google/pubs/pub40801/) (Dean & Barroso, 2013): fire a request, and if it doesn't return within your P50 latency, fire the same request in parallel. First response wins.
+
+The intuition: if 5% of requests spike due to independent random events, two parallel requests means only 0.25% of pairs spike on *both*. The tail collapses.
+
+**The surprise: hedging against the same upstream works.** HTTP/2 multiplexes streams — two `send_request()` calls on one connection become independent h2 streams. If one stalls in the dispatch channel, the other keeps making progress.
+
+```rust
+pub async fn forward_with_hedging_raw(
+    wire: &[u8],
+    primary: &Upstream,
+    secondary: &Upstream,
+    hedge_delay: Duration,
+    timeout_duration: Duration,
+) -> Result<Vec<u8>> {
+    let primary_fut = forward_query_raw(wire, primary, timeout_duration);
+    tokio::pin!(primary_fut);
+    let delay = sleep(hedge_delay);
+    tokio::pin!(delay);
+
+    // Phase 1: wait for primary to return OR the hedge delay.
+    tokio::select! {
+        result = &mut primary_fut => return result,
+        _ = &mut delay => {}
+    }
+
+    // Phase 2: hedge delay expired — fire secondary, keep primary alive.
+    let secondary_fut = forward_query_raw(wire, secondary, timeout_duration);
+    tokio::pin!(secondary_fut);
+
+    // First successful response wins.
+    tokio::select! {
+        r = primary_fut => r,
+        r = secondary_fut => r,
+    }
+}
+```
+
+The [production version](https://github.com/razvandimescu/numa/blob/main/src/forward.rs#L267) adds error handling — if one leg fails, it waits for the other. In production, Numa passes the same `&Upstream` twice when only one is configured. I extended hedging to all protocols — UDP (rescues packet loss on WiFi), DoT (rescues TLS handshake stalls). Configurable via `hedge_ms`; set to 0 to disable.
+
+**Caveat: hedging hurts on degraded networks.** When latency is consistently high (no random spikes, just slow), the hedge adds overhead with nothing to rescue. Hedging is a variance reducer, not a latency reducer — it only helps when spikes are *random*.
+
+---
+
+## Forwarding results
+
+5 iterations × 101 domains × 10 rounds, 5,050 samples per method. Hickory-resolver included as a reference (it uses h2 directly, no dispatch channel):
+
+| | Single | **Hedged** | Hickory (ref) |
+|---|---|---|---|
+| mean | 17.4ms | **14.3ms** | 16.8ms |
+| median | 10.4ms | **10.2ms** | 13.3ms |
+| p95 | 52.5ms | **28.6ms** | 37.7ms |
+| p99 | 113.4ms | **71.3ms** | 98.1ms |
+| σ | 30.6ms | **13.2ms** | 19.1ms |
+
+The internal improvement: hedging cut p95 by 45%, p99 by 37%, σ by 57%. The exact margin vs hickory varies with network conditions; the σ reduction is consistent across runs.
+
+## Recursive resolution: from 2.3 seconds to 538ms
+
+Forwarding is one job. Recursive resolution — walking from root hints through TLD nameservers to the authoritative server — is a different one. I started 15× behind Unbound on cold recursive p99 and traced it to four root causes.
+
+**1. Missing NS delegation caching.** I cached glue records (ns1's IP) but not the delegation itself. Every `.com` query walked from root. Fix: cache NS records from referral authority sections. (10 lines)
+
+**2. Expired cache entries caused full cold resolutions.** Fix: serve-stale ([RFC 8767](https://www.rfc-editor.org/rfc/rfc8767)) — return expired entries with TTL=1 while revalidating in the background. (20 lines)
+
+**3. Wasting 1,900ms per unreachable server.** 800ms UDP timeout + unconditional 1,500ms TCP fallback. Fix: 400ms UDP, TCP only for truncation. (5 lines)
+
+**4. Sequential NS queries on cold starts.** Fix: fire to the top 2 nameservers simultaneously. First response wins, SRTT recorded for both. Same hedging principle. (50 lines)
+
+<div class="before-after">
+<div class="ba-item">
+<div class="ba-label">p99 before</div>
+<div class="ba-value ba-before">2,367ms</div>
+</div>
+<div class="ba-arrow">&#8594;</div>
+<div class="ba-item">
+<div class="ba-label">p99 after</div>
+<div class="ba-value ba-after">538ms</div>
+</div>
+<div class="ba-item ba-ref">
+<div class="ba-label">Unbound (ref)</div>
+<div class="ba-value">748ms</div>
+</div>
+</div>
+
+Genuine cold benchmarks — unique subdomains, 1 query per domain, 5 iterations, 505 samples per server:
+
+| | Baseline | Final | Unbound (ref) |
+|---|---|---|---|
+| p99 | 2,367ms | **538ms** | 748ms |
+| σ | 254ms | **114ms** | 457ms |
+| median | — | 77.6ms | 74.7ms |
+
+Unbound wins median by ~4%. Where hedging shines is the tail — domains with slow or unreachable nameservers, where parallel queries turn worst-case sequential timeouts into races. Cache hits are tied at 0.1ms across Numa, Unbound, and AdGuard Home.
+
+What I'm exploring next: persistent SRTT data across restarts (currently cold-starts lose all server timing), aggressive NSEC caching to shortcut negative lookups, and adaptive hedge delays that tune themselves to observed network conditions instead of a fixed 10ms.
+
+---
+
+## Takeaways
+
+The real hero of this post is Dean & Barroso. Hedging works because **spikes are random, and two random draws rarely both lose**. It's effective for any HTTP/2 client, any language, any forwarder topology. Nobody we know of ships it by default.
+
+If you're building a Rust service that makes many small HTTP/2 requests to the same backend: check your flow control window sizes first, then implement hedging. Don't rewrite the client.
+
+Benchmarks are in [`benches/recursive_compare.rs`](https://github.com/razvandimescu/numa/blob/main/benches/recursive_compare.rs) — run them yourself. If you're using reqwest for tiny-payload workloads and try the window size fix, I'd love to hear if you see the same improvement.
+
+---
+
+Numa is a DNS resolver that runs on your laptop or phone. DoH, DoT, .numa local domains, ad blocking, developer overrides, a REST API, and all the optimization work in this post. [github.com/razvandimescu/numa](https://github.com/razvandimescu/numa).

numa.toml

@@ -58,6 +58,14 @@ api_port = 5380
 # [[forwarding]]
 # suffix = ["home.local", "home.arpa"]             # multiple suffixes → same upstream
 # upstream = "10.0.0.1"                            # port 53 default
+#
+# [[forwarding]]                                    # DoT upstream: tls://IP[:port]#hostname
+# suffix = ["google.com", "goog"]                   # hostname is the TLS SNI / cert name
+# upstream = "tls://9.9.9.9#dns.quad9.net"          # port 853 default
+#
+# [[forwarding]]                                    # DoH upstream: full https:// URL
+# suffix = "example.corp"
+# upstream = "https://dns.quad9.net/dns-query"
 
 # [blocking]
 # enabled = true               # set to false to disable ad blocking

scripts/generate-blog-index.sh

@@ -0,0 +1,239 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# Generate site/blog/index.html from blog/*.md frontmatter.
+# Reads title, description, date from YAML frontmatter in each post.
+# Sorts newest first (by date string — "April 2026" > "March 2026").
+
+OUT="site/blog/index.html"
+
+# Extract frontmatter fields from a markdown file
+extract() {
+  local file="$1" field="$2"
+  sed -n '/^---$/,/^---$/p' "$file" | grep "^${field}:" | sed "s/^${field}: *//"
+}
+
+# Collect posts: "date|name|title|description" per line
+posts=""
+sources="blog/*.md"
+if [ "${BLOG_INCLUDE_DRAFTS:-}" = "1" ] && ls drafts/*.md >/dev/null 2>&1; then
+  sources="blog/*.md drafts/*.md"
+fi
+for f in $sources; do
+  name=$(basename "$f" .md)
+  title=$(extract "$f" title)
+  desc=$(extract "$f" description)
+  date=$(extract "$f" date)
+  posts+="${date}|${name}|${title}|${desc}"$'\n'
+done
+
+# Sort by ISO date (YYYY-MM-DD), newest first
+posts=$(echo "$posts" | grep -v '^$' | sort -t'|' -k1 -r)
+
+# Format ISO date (YYYY-MM-DD) to "Month YYYY"
+format_date() {
+  local months=(January February March April May June July August September October November December)
+  local y="${1%%-*}"
+  local m="${1#*-}"; m="${m%%-*}"; m=$((10#$m))
+  echo "${months[$((m-1))]} $y"
+}
+
+# Generate post list items
+items=""
+while IFS='|' read -r date name title desc; do
+  display_date=$(format_date "$date")
+  items+="    <li>
+      <a href=\"/blog/posts/${name}.html\">
+        <div class=\"post-title\">${title}</div>
+        <div class=\"post-desc\">${desc}</div>
+        <div class=\"post-date\">${display_date}</div>
+      </a>
+    </li>
+"
+done <<< "$posts"
+
+# Write the full index.html — style matches the existing hand-maintained version
+cat > "$OUT" << HTMLEOF
+<!DOCTYPE html>
+<html lang="en">
+<head>
+<meta charset="UTF-8">
+<meta name="viewport" content="width=device-width, initial-scale=1.0">
+<title>Blog — Numa</title>
+<meta name="description" content="Technical writing about DNS, Rust, and building infrastructure from scratch.">
+<link rel="stylesheet" href="/fonts/fonts.css">
+<style>
+*, *::before, *::after { margin: 0; padding: 0; box-sizing: border-box; }
+
+:root {
+  --bg-deep: #f5f0e8;
+  --bg-surface: #ece5da;
+  --bg-card: #faf7f2;
+  --amber: #c0623a;
+  --amber-dim: #9e4e2d;
+  --teal: #6b7c4e;
+  --text-primary: #2c2418;
+  --text-secondary: #6b5e4f;
+  --text-dim: #a39888;
+  --border: rgba(0, 0, 0, 0.08);
+  --font-display: 'Instrument Serif', Georgia, serif;
+  --font-body: 'DM Sans', system-ui, sans-serif;
+  --font-mono: 'JetBrains Mono', monospace;
+}
+
+body {
+  background: var(--bg-deep);
+  color: var(--text-primary);
+  font-family: var(--font-body);
+  font-weight: 400;
+  line-height: 1.7;
+  -webkit-font-smoothing: antialiased;
+}
+
+body::before {
+  content: '';
+  position: fixed;
+  inset: 0;
+  background-image: url("data:image/svg+xml,%3Csvg viewBox='0 0 256 256' xmlns='http://www.w3.org/2000/svg'%3E%3Cfilter id='n'%3E%3CfeTurbulence type='fractalNoise' baseFrequency='0.9' numOctaves='4' stitchTiles='stitch'/%3E%3C/filter%3E%3Crect width='100%25' height='100%25' filter='url(%23n)' opacity='0.025'/%3E%3C/svg%3E");
+  pointer-events: none;
+  z-index: 9999;
+}
+
+.blog-nav {
+  padding: 1.5rem 2rem;
+  display: flex;
+  align-items: center;
+  gap: 1.5rem;
+}
+
+.blog-nav a {
+  font-family: var(--font-mono);
+  font-size: 0.75rem;
+  letter-spacing: 0.08em;
+  text-transform: uppercase;
+  color: var(--text-dim);
+  text-decoration: none;
+  transition: color 0.2s;
+}
+.blog-nav a:hover { color: var(--amber); }
+
+.blog-nav .wordmark {
+  font-family: var(--font-display);
+  font-size: 1.4rem;
+  font-weight: 400;
+  color: var(--text-primary);
+  text-decoration: none;
+  text-transform: none;
+  letter-spacing: -0.02em;
+}
+.blog-nav .wordmark:hover { color: var(--amber); }
+
+.blog-nav .sep {
+  color: var(--text-dim);
+  font-family: var(--font-mono);
+  font-size: 0.75rem;
+}
+
+.blog-index {
+  max-width: 720px;
+  margin: 0 auto;
+  padding: 3rem 2rem 6rem;
+}
+
+.blog-index h1 {
+  font-family: var(--font-display);
+  font-weight: 400;
+  font-size: 2.5rem;
+  margin-bottom: 3rem;
+}
+
+.post-list {
+  list-style: none;
+}
+
+.post-list li {
+  padding: 1.5rem 0;
+  border-bottom: 1px solid var(--border);
+}
+
+.post-list li:first-child {
+  border-top: 1px solid var(--border);
+}
+
+.post-list a {
+  text-decoration: none;
+  display: block;
+}
+
+.post-list .post-title {
+  font-family: var(--font-display);
+  font-size: 1.4rem;
+  font-weight: 600;
+  color: var(--text-primary);
+  line-height: 1.3;
+  margin-bottom: 0.4rem;
+  transition: color 0.2s;
+}
+
+.post-list a:hover .post-title {
+  color: var(--amber);
+}
+
+.post-list .post-desc {
+  font-size: 0.95rem;
+  color: var(--text-secondary);
+  line-height: 1.5;
+  margin-bottom: 0.4rem;
+}
+
+.post-list .post-date {
+  font-family: var(--font-mono);
+  font-size: 0.72rem;
+  color: var(--text-dim);
+  letter-spacing: 0.04em;
+}
+
+.blog-footer {
+  text-align: center;
+  padding: 3rem 2rem;
+  border-top: 1px solid var(--border);
+  max-width: 720px;
+  margin: 0 auto;
+}
+
+.blog-footer a {
+  font-family: var(--font-mono);
+  font-size: 0.75rem;
+  letter-spacing: 0.08em;
+  text-transform: uppercase;
+  color: var(--text-dim);
+  text-decoration: none;
+  margin: 0 1rem;
+}
+.blog-footer a:hover { color: var(--amber); }
+</style>
+</head>
+<body>
+
+<nav class="blog-nav">
+  <a href="/" class="wordmark">Numa</a>
+  <span class="sep">/</span>
+  <a href="/blog/">Blog</a>
+</nav>
+
+<main class="blog-index">
+  <h1>Blog</h1>
+  <ul class="post-list">
+${items}  </ul>
+</main>
+
+<footer class="blog-footer">
+  <a href="https://github.com/razvandimescu/numa">GitHub</a>
+  <a href="/">Home</a>
+</footer>
+
+</body>
+</html>
+HTMLEOF
+
+echo "  blog/index.html generated ($(echo "$posts" | wc -l | tr -d ' ') posts)"

scripts/serve-site.sh

@@ -0,0 +1,14 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+PORT="${1:-9000}"
+
+if [[ "${1:-}" == "--drafts" ]] || [[ "${2:-}" == "--drafts" ]]; then
+  PORT="${PORT//--drafts/9000}"  # default port if --drafts was first arg
+  make blog-drafts
+else
+  make blog
+fi
+
+echo "Serving site at http://localhost:$PORT"
+cd site && python3 -m http.server "$PORT"

site/blog-template.html

@@ -267,9 +267,105 @@ body::before {
 .blog-footer a:hover { color: var(--amber); }
 
 /* --- Responsive --- */
+/* Hero metrics cards */
+.hero-metrics {
+  display: grid;
+  grid-template-columns: repeat(3, 1fr);
+  gap: 1rem;
+  margin: 2rem 0;
+}
+.metric-card {
+  background: var(--bg-card);
+  border: 1px solid var(--border);
+  border-radius: 6px;
+  padding: 1.25rem;
+  text-align: center;
+}
+.metric-vs {
+  font-family: var(--font-mono);
+  font-size: 0.7rem;
+  letter-spacing: 0.08em;
+  text-transform: uppercase;
+  color: var(--text-dim);
+  margin-bottom: 0.5rem;
+}
+.metric-value {
+  font-family: var(--font-display);
+  font-size: 2.4rem;
+  font-weight: 400;
+  color: var(--amber);
+  line-height: 1.1;
+}
+.metric-label {
+  font-size: 0.82rem;
+  color: var(--text-secondary);
+  margin-top: 0.5rem;
+  line-height: 1.3;
+}
+
+/* Before/after progression */
+.before-after {
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  gap: 1.5rem;
+  margin: 2rem 0;
+  padding: 1.5rem;
+  background: var(--bg-card);
+  border: 1px solid var(--border);
+  border-radius: 6px;
+}
+.ba-item { text-align: center; }
+.ba-label {
+  font-family: var(--font-mono);
+  font-size: 0.7rem;
+  letter-spacing: 0.08em;
+  text-transform: uppercase;
+  color: var(--text-dim);
+  margin-bottom: 0.3rem;
+}
+.ba-value {
+  font-family: var(--font-display);
+  font-size: 1.8rem;
+  font-weight: 400;
+  color: var(--text-secondary);
+}
+.ba-before {
+  text-decoration: line-through;
+  text-decoration-color: rgba(192, 98, 58, 0.4);
+  color: var(--text-dim);
+}
+.ba-after { color: var(--amber); }
+.ba-arrow { font-size: 1.5rem; color: var(--text-dim); }
+.ba-ref {
+  border-left: 1px solid var(--border);
+  padding-left: 1.5rem;
+}
+
+/* Spike highlight */
+.spike {
+  background: rgba(192, 98, 58, 0.12);
+  padding: 0.15em 0.5em;
+  border-radius: 3px;
+  font-weight: 600;
+  color: var(--amber-dim);
+}
+
+/* Section dividers */
+.article hr {
+  border: none;
+  height: 1px;
+  background: var(--border);
+  margin: 3rem auto;
+  max-width: 120px;
+}
+
 @media (max-width: 640px) {
   .article { padding: 2rem 1.25rem 4rem; }
   .article pre { padding: 1rem; margin-left: -0.5rem; margin-right: -0.5rem; border-radius: 0; border-left: none; border-right: none; }
+  .hero-metrics { grid-template-columns: 1fr; }
+  .before-after { flex-direction: column; gap: 0.75rem; }
+  .ba-ref { border-left: none; border-top: 1px solid var(--border); padding-left: 0; padding-top: 0.75rem; }
 }
 </style>
 </head>

site/blog/index.html

@@ -168,10 +168,17 @@ body::before {
 <main class="blog-index">
   <h1>Blog</h1>
   <ul class="post-list">
+    <li>
+      <a href="/blog/posts/fixing-doh-tail-latency.html">
+        <div class="post-title">Fixing DNS tail latency with a 5-line config and a 50-line function</div>
+        <div class="post-desc">Periodic 40-140ms DoH spikes from hyper's dispatch channel. The fix was reqwest window tuning and request hedging — Dean & Barroso's "The Tail at Scale," applied to a DNS forwarder. Same ideas took cold recursive p99 from 2.3 seconds to 538ms.</div>
+        <div class="post-date">April 2026</div>
+      </a>
+    </li>
     <li>
       <a href="/blog/posts/dot-from-scratch.html">
         <div class="post-title">DNS-over-TLS from Scratch in Rust</div>
-        <div class="post-desc">Building RFC 7858 on top of rustls — length-prefix framing, ALPN cross-protocol defense, iPhone dogfooding, and two bugs that only the strict clients caught.</div>
+        <div class="post-desc">Building RFC 7858 on top of rustls — length-prefix framing, ALPN cross-protocol defense, and two bugs that only the strict clients caught.</div>
         <div class="post-date">April 2026</div>
       </a>
     </li>
@@ -185,7 +192,7 @@ body::before {
     <li>
       <a href="/blog/posts/dns-from-scratch.html">
         <div class="post-title">I Built a DNS Resolver from Scratch in Rust</div>
-        <div class="post-desc">How DNS actually works at the wire level — label compression, TTL tricks, DoH implementation, and what I learned building a resolver with zero DNS libraries.</div>
+        <div class="post-desc">How DNS actually works at the wire level — label compression, TTL tricks, DoH, and what surprised me building a resolver with zero DNS libraries.</div>
         <div class="post-date">March 2026</div>
       </a>
     </li>

src/api.rs

@@ -1020,53 +1020,10 @@ mod tests {
     use super::*;
     use axum::body::Body;
     use http::Request;
-    use std::sync::{Mutex, RwLock};
     use tower::ServiceExt;
 
     async fn test_ctx() -> Arc<ServerCtx> {
-        let socket = tokio::net::UdpSocket::bind("127.0.0.1:0").await.unwrap();
+        Arc::new(crate::testutil::test_ctx().await)
-        Arc::new(ServerCtx {
-            socket,
-            zone_map: std::collections::HashMap::new(),
-            cache: RwLock::new(crate::cache::DnsCache::new(100, 60, 86400)),
-            refreshing: Mutex::new(std::collections::HashSet::new()),
-            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_pool: Mutex::new(crate::forward::UpstreamPool::new(
-                vec![crate::forward::Upstream::Udp(
-                    "127.0.0.1:53".parse().unwrap(),
-                )],
-                vec![],
-            )),
-            upstream_auto: false,
-            upstream_port: 53,
-            lan_ip: Mutex::new(std::net::Ipv4Addr::LOCALHOST),
-            timeout: std::time::Duration::from_secs(3),
-            hedge_delay: std::time::Duration::ZERO,
-            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::Forward,
-            root_hints: Vec::new(),
-            srtt: RwLock::new(crate::srtt::SrttCache::new(true)),
-            inflight: Mutex::new(std::collections::HashMap::new()),
-            dnssec_enabled: false,
-            dnssec_strict: false,
-            health_meta: crate::health::HealthMeta::test_fixture(),
-            ca_pem: None,
-            mobile_enabled: false,
-            mobile_port: 8765,
-        })
     }
 
     #[tokio::test]

src/config.rs

@@ -46,12 +46,12 @@ pub struct ForwardingRuleConfig {
 
 impl ForwardingRuleConfig {
     fn to_runtime_rules(&self) -> Result<Vec<crate::system_dns::ForwardingRule>> {
-        let addr = crate::forward::parse_upstream_addr(&self.upstream, 53)
+        let upstream = crate::forward::parse_upstream(&self.upstream, 53)
             .map_err(|e| format!("forwarding rule for upstream '{}': {}", self.upstream, e))?;
         Ok(self
             .suffix
             .iter()
-            .map(|s| crate::system_dns::ForwardingRule::new(s.clone(), addr))
+            .map(|s| crate::system_dns::ForwardingRule::new(s.clone(), upstream.clone()))
             .collect())
     }
 }
@@ -710,6 +710,10 @@ mod tests {
         };
         let runtime = rule.to_runtime_rules().unwrap();
         assert_eq!(runtime.len(), 1);
+        assert!(matches!(
+            runtime[0].upstream,
+            crate::forward::Upstream::Udp(_)
+        ));
         assert_eq!(runtime[0].upstream.to_string(), "100.90.1.63:5361");
         assert_eq!(runtime[0].suffix, "home.local");
     }
@@ -733,6 +737,38 @@ mod tests {
         assert!(rule.to_runtime_rules().is_err());
     }
 
+    #[test]
+    fn forwarding_upstream_accepts_dot_scheme() {
+        let rule = ForwardingRuleConfig {
+            suffix: vec!["google.com".to_string()],
+            upstream: "tls://9.9.9.9#dns.quad9.net".to_string(),
+        };
+        let runtime = rule
+            .to_runtime_rules()
+            .expect("tls:// upstream should parse");
+        assert_eq!(runtime.len(), 1);
+        assert_eq!(
+            runtime[0].upstream.to_string(),
+            "tls://9.9.9.9:853#dns.quad9.net"
+        );
+    }
+
+    #[test]
+    fn forwarding_upstream_accepts_doh_scheme() {
+        let rule = ForwardingRuleConfig {
+            suffix: vec!["goog".to_string()],
+            upstream: "https://dns.quad9.net/dns-query".to_string(),
+        };
+        let runtime = rule
+            .to_runtime_rules()
+            .expect("https:// upstream should parse");
+        assert_eq!(runtime.len(), 1);
+        assert_eq!(
+            runtime[0].upstream.to_string(),
+            "https://dns.quad9.net/dns-query"
+        );
+    }
+
     #[test]
     fn forwarding_config_rules_take_precedence_over_discovered() {
         let config_rules = vec![ForwardingRuleConfig {
@@ -741,7 +777,7 @@ mod tests {
         }];
         let discovered = vec![crate::system_dns::ForwardingRule::new(
             "home.local".to_string(),
-            "192.168.1.1:53".parse().unwrap(),
+            crate::forward::Upstream::Udp("192.168.1.1:53".parse().unwrap()),
         )];
         let merged = merge_forwarding_rules(&config_rules, discovered).unwrap();
         let picked = crate::system_dns::match_forwarding_rule("host.home.local", &merged)
@@ -757,7 +793,7 @@ mod tests {
         }];
         let discovered = vec![crate::system_dns::ForwardingRule::new(
             "corp.example".to_string(),
-            "192.168.1.1:53".parse().unwrap(),
+            crate::forward::Upstream::Udp("192.168.1.1:53".parse().unwrap()),
         )];
         let merged = merge_forwarding_rules(&config_rules, discovered).unwrap();
         assert_eq!(merged.len(), 2);

src/ctx.rs

@@ -88,7 +88,7 @@ pub async fn resolve_query(
     src_addr: SocketAddr,
     ctx: &Arc<ServerCtx>,
     transport: Transport,
-) -> crate::Result<BytePacketBuffer> {
+) -> crate::Result<(BytePacketBuffer, QueryPath)> {
     let start = Instant::now();
 
     let (qname, qtype) = match query.questions.first() {
@@ -96,7 +96,8 @@ pub async fn resolve_query(
         None => return Err("empty question section".into()),
     };
 
-    // Pipeline: overrides -> .tld interception -> blocklist -> local zones -> cache -> upstream
+    // Pipeline: overrides -> .localhost -> local zones -> special-use (unless forwarded)
+    //        -> .tld proxy -> blocklist -> cache -> forwarding -> recursive/upstream
     // Each lock is scoped to avoid holding MutexGuard across await points.
     let (response, path, dnssec) = {
         let override_record = ctx.overrides.read().unwrap().lookup(&qname);
@@ -119,8 +120,10 @@ pub async fn resolve_query(
             let mut resp = DnsPacket::response_from(&query, ResultCode::NOERROR);
             resp.answers = records.clone();
             (resp, QueryPath::Local, DnssecStatus::Indeterminate)
-        } else if is_special_use_domain(&qname) {
+        } else if is_special_use_domain(&qname)
-            // RFC 6761/8880: private PTR, DDR, NAT64 — answer locally
+            && crate::system_dns::match_forwarding_rule(&qname, &ctx.forwarding_rules).is_none()
+        {
+            // RFC 6761/8880: answer locally unless a forwarding rule covers this zone.
             let resp = special_use_response(&query, &qname, qtype);
             (resp, QueryPath::Local, DnssecStatus::Indeterminate)
         } else if !ctx.proxy_tld_suffix.is_empty()
@@ -187,13 +190,12 @@ pub async fn resolve_query(
                     resp.header.authed_data = true;
                 }
                 (resp, QueryPath::Cached, cached_dnssec)
-            } else if let Some(fwd_addr) =
+            } else if let Some(upstream) =
                 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_and_cache(raw_wire, upstream, ctx, &qname, qtype).await {
-                match forward_and_cache(raw_wire, &upstream, ctx, &qname, qtype).await {
                     Ok(resp) => (resp, QueryPath::Forwarded, DnssecStatus::Indeterminate),
                     Err(e) => {
                         error!(
@@ -373,7 +375,7 @@ pub async fn resolve_query(
         dnssec,
     });
 
-    Ok(resp_buffer)
+    Ok((resp_buffer, path))
 }
 
 fn cache_and_parse(
@@ -457,7 +459,7 @@ pub async fn handle_query(
         }
     };
     match resolve_query(query, &buffer.buf[..raw_len], src_addr, ctx, transport).await {
-        Ok(resp_buffer) => {
+        Ok((resp_buffer, _)) => {
             ctx.socket.send_to(resp_buffer.filled(), src_addr).await?;
         }
         Err(e) => {
@@ -1036,4 +1038,219 @@ mod tests {
             "error message must be preserved for logging"
         );
     }
+
+    // ---- Full-pipeline resolve_query tests ----
+
+    /// Send a query through the full resolve_query pipeline and return
+    /// the parsed response + query path.
+    async fn resolve_in_test(
+        ctx: &Arc<ServerCtx>,
+        domain: &str,
+        qtype: QueryType,
+    ) -> (DnsPacket, QueryPath) {
+        let query = DnsPacket::query(0xBEEF, domain, qtype);
+        let mut buf = BytePacketBuffer::new();
+        query.write(&mut buf).unwrap();
+        let raw = &buf.buf[..buf.pos];
+        let src: SocketAddr = "127.0.0.1:1234".parse().unwrap();
+
+        let (resp_buf, path) = resolve_query(query, raw, src, ctx, Transport::Udp)
+            .await
+            .unwrap();
+
+        let mut resp_parse_buf = BytePacketBuffer::from_bytes(resp_buf.filled());
+        let resp = DnsPacket::from_buffer(&mut resp_parse_buf).unwrap();
+        (resp, path)
+    }
+
+    #[tokio::test]
+    async fn special_use_private_ptr_returns_nxdomain() {
+        let ctx = Arc::new(crate::testutil::test_ctx().await);
+        let (resp, path) =
+            resolve_in_test(&ctx, "153.188.168.192.in-addr.arpa", QueryType::PTR).await;
+        assert_eq!(path, QueryPath::Local);
+        assert_eq!(resp.header.rescode, ResultCode::NXDOMAIN);
+    }
+
+    #[tokio::test]
+    async fn forwarding_rule_overrides_special_use_domain() {
+        let mut resp = DnsPacket::new();
+        resp.header.response = true;
+        resp.header.rescode = ResultCode::NOERROR;
+        let upstream_addr = crate::testutil::mock_upstream(resp).await;
+
+        let mut ctx = crate::testutil::test_ctx().await;
+        ctx.forwarding_rules = vec![ForwardingRule::new(
+            "168.192.in-addr.arpa".to_string(),
+            Upstream::Udp(upstream_addr),
+        )];
+        let ctx = Arc::new(ctx);
+
+        let (resp, path) =
+            resolve_in_test(&ctx, "153.188.168.192.in-addr.arpa", QueryType::PTR).await;
+
+        assert_eq!(
+            path,
+            QueryPath::Forwarded,
+            "forwarding rule must take precedence over special-use NXDOMAIN"
+        );
+        assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+    }
+
+    #[tokio::test]
+    async fn pipeline_override_takes_precedence() {
+        let ctx = crate::testutil::test_ctx().await;
+        ctx.overrides
+            .write()
+            .unwrap()
+            .insert("override.test", "1.2.3.4", 60, None)
+            .unwrap();
+        let ctx = Arc::new(ctx);
+
+        let (resp, path) = resolve_in_test(&ctx, "override.test", QueryType::A).await;
+        assert_eq!(path, QueryPath::Overridden);
+        assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+        assert_eq!(resp.answers.len(), 1);
+    }
+
+    #[tokio::test]
+    async fn pipeline_localhost_resolves_to_loopback() {
+        let ctx = Arc::new(crate::testutil::test_ctx().await);
+
+        let (resp, path) = resolve_in_test(&ctx, "localhost", QueryType::A).await;
+        assert_eq!(path, QueryPath::Local);
+        assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+        match &resp.answers[0] {
+            DnsRecord::A { addr, .. } => assert_eq!(*addr, Ipv4Addr::LOCALHOST),
+            other => panic!("expected A record, got {:?}", other),
+        }
+    }
+
+    #[tokio::test]
+    async fn pipeline_localhost_subdomain_resolves_to_loopback() {
+        let ctx = Arc::new(crate::testutil::test_ctx().await);
+
+        let (resp, path) = resolve_in_test(&ctx, "app.localhost", QueryType::A).await;
+        assert_eq!(path, QueryPath::Local);
+        assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+        match &resp.answers[0] {
+            DnsRecord::A { addr, .. } => assert_eq!(*addr, Ipv4Addr::LOCALHOST),
+            other => panic!("expected A record, got {:?}", other),
+        }
+    }
+
+    #[tokio::test]
+    async fn pipeline_local_zone_returns_configured_record() {
+        let mut ctx = crate::testutil::test_ctx().await;
+        let mut inner = HashMap::new();
+        inner.insert(
+            QueryType::A,
+            vec![DnsRecord::A {
+                domain: "myapp.test".to_string(),
+                addr: Ipv4Addr::new(10, 0, 0, 42),
+                ttl: 300,
+            }],
+        );
+        ctx.zone_map.insert("myapp.test".to_string(), inner);
+        let ctx = Arc::new(ctx);
+
+        let (resp, path) = resolve_in_test(&ctx, "myapp.test", QueryType::A).await;
+        assert_eq!(path, QueryPath::Local);
+        assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+        match &resp.answers[0] {
+            DnsRecord::A { addr, .. } => assert_eq!(*addr, Ipv4Addr::new(10, 0, 0, 42)),
+            other => panic!("expected A record, got {:?}", other),
+        }
+    }
+
+    #[tokio::test]
+    async fn pipeline_tld_proxy_resolves_service() {
+        let ctx = crate::testutil::test_ctx().await;
+        ctx.services.lock().unwrap().insert("grafana", 3000);
+        let ctx = Arc::new(ctx);
+
+        let (resp, path) = resolve_in_test(&ctx, "grafana.numa", QueryType::A).await;
+        assert_eq!(path, QueryPath::Local);
+        assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+        match &resp.answers[0] {
+            DnsRecord::A { addr, .. } => assert_eq!(*addr, Ipv4Addr::LOCALHOST),
+            other => panic!("expected A record, got {:?}", other),
+        }
+    }
+
+    #[tokio::test]
+    async fn pipeline_blocklist_sinkhole() {
+        let ctx = crate::testutil::test_ctx().await;
+        let mut domains = std::collections::HashSet::new();
+        domains.insert("ads.tracker.test".to_string());
+        ctx.blocklist.write().unwrap().swap_domains(domains, vec![]);
+        let ctx = Arc::new(ctx);
+
+        let (resp, path) = resolve_in_test(&ctx, "ads.tracker.test", QueryType::A).await;
+        assert_eq!(path, QueryPath::Blocked);
+        assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+        match &resp.answers[0] {
+            DnsRecord::A { addr, .. } => assert_eq!(*addr, Ipv4Addr::UNSPECIFIED),
+            other => panic!("expected sinkhole A record, got {:?}", other),
+        }
+    }
+
+    #[tokio::test]
+    async fn pipeline_cache_hit() {
+        let ctx = Arc::new(crate::testutil::test_ctx().await);
+
+        // Pre-populate cache with a response
+        let mut pkt = DnsPacket::new();
+        pkt.header.response = true;
+        pkt.header.rescode = ResultCode::NOERROR;
+        pkt.questions.push(crate::question::DnsQuestion {
+            name: "cached.test".to_string(),
+            qtype: QueryType::A,
+        });
+        pkt.answers.push(DnsRecord::A {
+            domain: "cached.test".to_string(),
+            addr: Ipv4Addr::new(5, 5, 5, 5),
+            ttl: 3600,
+        });
+        ctx.cache
+            .write()
+            .unwrap()
+            .insert("cached.test", QueryType::A, &pkt);
+
+        let (resp, path) = resolve_in_test(&ctx, "cached.test", QueryType::A).await;
+        assert_eq!(path, QueryPath::Cached);
+        assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+    }
+
+    #[tokio::test]
+    async fn pipeline_forwarding_returns_upstream_answer() {
+        let mut upstream_resp = DnsPacket::new();
+        upstream_resp.header.response = true;
+        upstream_resp.header.rescode = ResultCode::NOERROR;
+        upstream_resp.answers.push(DnsRecord::A {
+            domain: "internal.corp".to_string(),
+            addr: Ipv4Addr::new(10, 1, 2, 3),
+            ttl: 600,
+        });
+        let upstream_addr = crate::testutil::mock_upstream(upstream_resp).await;
+
+        let mut ctx = crate::testutil::test_ctx().await;
+        ctx.forwarding_rules = vec![ForwardingRule::new(
+            "corp".to_string(),
+            Upstream::Udp(upstream_addr),
+        )];
+        let ctx = Arc::new(ctx);
+
+        let (resp, path) = resolve_in_test(&ctx, "internal.corp", QueryType::A).await;
+        assert_eq!(path, QueryPath::Forwarded);
+        assert_eq!(resp.header.rescode, ResultCode::NOERROR);
+        assert_eq!(resp.answers.len(), 1);
+        match &resp.answers[0] {
+            DnsRecord::A { domain, addr, .. } => {
+                assert_eq!(domain, "internal.corp");
+                assert_eq!(*addr, Ipv4Addr::new(10, 1, 2, 3));
+            }
+            other => panic!("expected A record, got {:?}", other),
+        }
+    }
 }

src/doh.rs

@@ -113,7 +113,7 @@ async fn resolve_doh(
     let questions = query.questions.clone();
 
     match resolve_query(query, dns_bytes, src, ctx, Transport::Doh).await {
-        Ok(resp_buffer) => {
+        Ok((resp_buffer, _)) => {
             let min_ttl = extract_min_ttl(resp_buffer.filled());
             dns_response(resp_buffer.filled(), min_ttl)
         }

src/dot.rs

@@ -211,7 +211,7 @@ async fn handle_dot_connection<S>(
         )
         .await
         {
-            Ok(resp_buffer) => {
+            Ok((resp_buffer, _)) => {
                 if write_framed(&mut stream, resp_buffer.filled())
                     .await
                     .is_err()
@@ -279,7 +279,7 @@ where
 mod tests {
     use super::*;
     use std::collections::HashMap;
-    use std::sync::{Mutex, RwLock};
+    use std::sync::Mutex;
 
     use rcgen::{CertificateParams, DnType, KeyPair};
     use rustls::pki_types::{CertificateDer, PrivateKeyDer, PrivatePkcs8KeyDer, ServerName};
@@ -344,63 +344,29 @@ mod tests {
     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();
+        let upstream_addr = crate::testutil::blackhole_upstream();
-        // 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 mut ctx = crate::testutil::test_ctx().await;
-        let upstream_addr = blackhole.local_addr().unwrap();
+        ctx.zone_map = {
-        let ctx = Arc::new(ServerCtx {
+            let mut m = HashMap::new();
-            socket,
+            let mut inner = HashMap::new();
-            zone_map: {
+            inner.insert(
-                let mut m = HashMap::new();
+                QueryType::A,
-                let mut inner = HashMap::new();
+                vec![DnsRecord::A {
-                inner.insert(
+                    domain: "dot-test.example".to_string(),
-                    QueryType::A,
+                    addr: std::net::Ipv4Addr::new(10, 0, 0, 1),
-                    vec![DnsRecord::A {
+                    ttl: 300,
-                        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
-                );
+        };
-                m.insert("dot-test.example".to_string(), inner);
+        ctx.upstream_pool = Mutex::new(crate::forward::UpstreamPool::new(
-                m
+            vec![crate::forward::Upstream::Udp(upstream_addr)],
-            },
+            vec![],
-            cache: RwLock::new(crate::cache::DnsCache::new(100, 60, 86400)),
+        ));
-            refreshing: Mutex::new(std::collections::HashSet::new()),
+        ctx.tls_config = Some(arc_swap::ArcSwap::from(server_tls));
-            stats: Mutex::new(crate::stats::ServerStats::new()),
+        let ctx = Arc::new(ctx);
-            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_pool: Mutex::new(crate::forward::UpstreamPool::new(
-                vec![crate::forward::Upstream::Udp(upstream_addr)],
-                vec![],
-            )),
-            upstream_auto: false,
-            upstream_port: 53,
-            lan_ip: Mutex::new(std::net::Ipv4Addr::LOCALHOST),
-            timeout: Duration::from_millis(200),
-            hedge_delay: Duration::ZERO,
-            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,
-            health_meta: crate::health::HealthMeta::test_fixture(),
-            ca_pem: None,
-            mobile_enabled: false,
-            mobile_port: 8765,
-        });
 
         let listener = TcpListener::bind("127.0.0.1:0").await.unwrap();
         let addr = listener.local_addr().unwrap();

src/forward.rs

@@ -36,6 +36,12 @@ impl PartialEq for Upstream {
     }
 }
 
+impl fmt::Debug for Upstream {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        fmt::Display::fmt(self, f)
+    }
+}
+
 impl fmt::Display for Upstream {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         match self {
@@ -49,7 +55,10 @@ impl fmt::Display for Upstream {
     }
 }
 
-pub fn parse_upstream_addr(s: &str, default_port: u16) -> std::result::Result<SocketAddr, String> {
+pub(crate) fn parse_upstream_addr(
+    s: &str,
+    default_port: u16,
+) -> std::result::Result<SocketAddr, String> {
     // Try full socket addr first: "1.2.3.4:5353" or "[::1]:5353"
     if let Ok(addr) = s.parse::<SocketAddr>() {
         return Ok(addr);

src/lib.rs

@@ -28,6 +28,9 @@ pub mod system_dns;
 pub mod tls;
 pub mod wire;
 
+#[cfg(test)]
+pub(crate) mod testutil;
+
 pub type Error = Box<dyn std::error::Error + Send + Sync>;
 pub type Result<T> = std::result::Result<T, Error>;
 

src/system_dns.rs

@@ -2,6 +2,8 @@ use std::net::SocketAddr;
 
 use log::info;
 
+use crate::forward::Upstream;
+
 fn print_recursive_hint() {
     let is_recursive = crate::config::load_config("numa.toml")
         .map(|c| c.config.upstream.mode == crate::config::UpstreamMode::Recursive)
@@ -22,11 +24,11 @@ fn is_loopback_or_stub(addr: &str) -> bool {
 pub struct ForwardingRule {
     pub suffix: String,
     dot_suffix: String, // pre-computed ".suffix" for zero-alloc matching
-    pub upstream: SocketAddr,
+    pub upstream: Upstream,
 }
 
 impl ForwardingRule {
-    pub fn new(suffix: String, upstream: SocketAddr) -> Self {
+    pub fn new(suffix: String, upstream: Upstream) -> Self {
         let dot_suffix = format!(".{}", suffix);
         Self {
             suffix,
@@ -233,7 +235,7 @@ fn discover_macos() -> SystemDnsInfo {
 #[cfg(any(target_os = "macos", target_os = "linux"))]
 fn make_rule(domain: &str, nameserver: &str) -> Option<ForwardingRule> {
     let addr = crate::forward::parse_upstream_addr(nameserver, 53).ok()?;
-    Some(ForwardingRule::new(domain.to_string(), addr))
+    Some(ForwardingRule::new(domain.to_string(), Upstream::Udp(addr)))
 }
 
 #[cfg(target_os = "linux")]
@@ -822,10 +824,13 @@ fn uninstall_windows() -> Result<(), String> {
 /// 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.
-pub fn match_forwarding_rule(domain: &str, rules: &[ForwardingRule]) -> Option<SocketAddr> {
+pub fn match_forwarding_rule<'a>(
+    domain: &str,
+    rules: &'a [ForwardingRule],
+) -> Option<&'a Upstream> {
     for rule in rules {
         if domain == rule.suffix || domain.ends_with(&rule.dot_suffix) {
-            return Some(rule.upstream);
+            return Some(&rule.upstream);
         }
     }
     None

src/testutil.rs

@@ -0,0 +1,95 @@
+use std::collections::{HashMap, HashSet};
+use std::net::{Ipv4Addr, SocketAddr};
+use std::path::PathBuf;
+use std::sync::{Mutex, RwLock};
+use std::time::Duration;
+
+use tokio::net::UdpSocket;
+
+use crate::blocklist::BlocklistStore;
+use crate::buffer::BytePacketBuffer;
+use crate::cache::DnsCache;
+use crate::config::UpstreamMode;
+use crate::ctx::ServerCtx;
+use crate::forward::{Upstream, UpstreamPool};
+use crate::health::HealthMeta;
+use crate::lan::PeerStore;
+use crate::override_store::OverrideStore;
+use crate::packet::DnsPacket;
+use crate::query_log::QueryLog;
+use crate::service_store::ServiceStore;
+use crate::srtt::SrttCache;
+use crate::stats::ServerStats;
+/// Minimal `ServerCtx` for tests. Override fields after construction
+/// (all fields are `pub`), then wrap in `Arc`.
+pub async fn test_ctx() -> ServerCtx {
+    let socket = UdpSocket::bind("127.0.0.1:0").await.unwrap();
+    ServerCtx {
+        socket,
+        zone_map: HashMap::new(),
+        cache: RwLock::new(DnsCache::new(100, 60, 86400)),
+        refreshing: Mutex::new(HashSet::new()),
+        stats: Mutex::new(ServerStats::new()),
+        overrides: RwLock::new(OverrideStore::new()),
+        blocklist: RwLock::new(BlocklistStore::new()),
+        query_log: Mutex::new(QueryLog::new(100)),
+        services: Mutex::new(ServiceStore::new()),
+        lan_peers: Mutex::new(PeerStore::new(90)),
+        forwarding_rules: Vec::new(),
+        upstream_pool: Mutex::new(UpstreamPool::new(
+            vec![Upstream::Udp("127.0.0.1:53".parse().unwrap())],
+            vec![],
+        )),
+        upstream_auto: false,
+        upstream_port: 53,
+        lan_ip: Mutex::new(Ipv4Addr::LOCALHOST),
+        timeout: Duration::from_millis(200),
+        hedge_delay: Duration::ZERO,
+        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: PathBuf::from("/tmp"),
+        data_dir: PathBuf::from("/tmp"),
+        tls_config: None,
+        upstream_mode: UpstreamMode::Forward,
+        root_hints: Vec::new(),
+        srtt: RwLock::new(SrttCache::new(true)),
+        inflight: Mutex::new(HashMap::new()),
+        dnssec_enabled: false,
+        dnssec_strict: false,
+        health_meta: HealthMeta::test_fixture(),
+        ca_pem: None,
+        mobile_enabled: false,
+        mobile_port: 8765,
+    }
+}
+
+/// Spawn a UDP socket that replies to the first DNS query with the given
+/// response packet (patching the query ID to match). Returns the socket address.
+pub async fn mock_upstream(response: DnsPacket) -> SocketAddr {
+    let sock = UdpSocket::bind("127.0.0.1:0").await.unwrap();
+    let addr = sock.local_addr().unwrap();
+    tokio::spawn(async move {
+        let mut buf = [0u8; 512];
+        let (_, src) = sock.recv_from(&mut buf).await.unwrap();
+        let query_id = u16::from_be_bytes([buf[0], buf[1]]);
+        let mut resp = response;
+        resp.header.id = query_id;
+        let mut out = BytePacketBuffer::new();
+        resp.write(&mut out).unwrap();
+        sock.send_to(out.filled(), src).await.unwrap();
+    });
+    addr
+}
+
+/// UDP socket that accepts connections but never replies.
+/// Useful as an upstream that triggers timeouts.
+pub fn blackhole_upstream() -> SocketAddr {
+    let sock = std::net::UdpSocket::bind("127.0.0.1:0").unwrap();
+    let addr = sock.local_addr().unwrap();
+    // Leak so it stays bound for the duration of the test process.
+    Box::leak(Box::new(sock));
+    addr
+}