//! The composition root: everything `nxdns run` owns, in the order it is built.
//!
//! Nothing else in the program constructs a collaborator. Each module takes its
//! dependencies as parameters and knows nothing about who supplies them, so
//! this file is the only place where the object graph exists — and the only
//! place a lifetime spans the whole process.
//!
//! Two rules shape it:
//!
//! 1. **Nothing that is pointed at may move.** `Pool.Entry.client` erases a
//!    pointer into a DoH or DoT client, the `Logger`'s queue holds waiting
//!    tasks in intrusive lists, and the `Manager` is addressed through `self`.
//!    Every such value therefore lives in `serve`'s frame or in an owned heap
//!    allocation, never in a temporary that is copied afterwards.
//! 2. **Every background loop is a task in one `std.Io.Group`.** Shutdown is
//!    `group.cancel`, which requests cancellation and joins, so no loop can
//!    still be running when the resources it borrows are released.
//!
//! Failure policy: a startup failure prints one line to the runner's error
//! writer and exits. A configuration the operator can fix exits 2, so that
//! `nxdns check` is the obvious next step; anything else exits 1. Once serving
//! starts, nothing is fatal — the pipeline answers queries with a snapshot or
//! without one (ruling 4), and every loop logs its own failures and keeps
//! going.

const std = @import("std");
const Allocator = std.mem.Allocator;
const Certificate = std.crypto.Certificate;
const Writer = std.Io.Writer;
const net = std.Io.net;
const tls = std.crypto.tls;

const api_limiter = @import("web/api_limiter.zig");
const auth = @import("web/auth.zig");
const bootstrap = @import("config/bootstrap.zig");
const cert_store = @import("server/cert_store.zig");
const cli = @import("cli.zig");
const clients = @import("server/clients.zig");
const config_export = @import("config/export.zig");
const db = @import("storage/db.zig");
const disk_monitor = @import("storage/disk_monitor.zig");
const dns_cache = @import("cache/dns_cache.zig");
const doh_client = @import("upstream/doh_client.zig");
const doh_server = @import("server/doh_server.zig");
const dot_client = @import("upstream/dot_client.zig");
const dot_server = @import("server/dot_server.zig");
const faults = @import("config/faults.zig");
const fetcher = @import("filter/fetcher.zig");
const forward_zones = @import("local/forward_zones.zig");
const handler = @import("server/handler.zig");
const http_util = @import("web/http_util.zig");
const local_records = @import("local/records.zig");
const local_tables = @import("server/local_tables.zig");
const logger_mod = @import("storage/logger.zig");
const logging = @import("platform/logging.zig");
const manager_mod = @import("filter/manager.zig");
const migrations = @import("storage/migrations.zig");
const model = @import("config/model.zig");
const pause = @import("server/pause.zig");
const pool_mod = @import("upstream/pool.zig");
const query_sink = @import("server/query_sink.zig");
const rate_limiter = @import("server/rate_limiter.zig");
const retention_mod = @import("storage/retention.zig");
const safe_url = @import("safe_url.zig");
const shutdown = @import("server/shutdown.zig");
const sse = @import("web/sse.zig");
const static = @import("web/static.zig");
const tcp_server = @import("server/tcp_server.zig");
const transport = @import("upstream/transport.zig");
const udp_server = @import("server/udp_server.zig");
const validate = @import("config/validate.zig");
const version = @import("version.zig");
const web_server = @import("web/server.zig");

const log = std.log.scoped(.nxdns);

/// How long the cache and rate-limiter sweeps wait between passes. Both tables
/// evict lazily on lookup as well; this is what keeps memory bounded for a
/// client that asked once and never came back.
const maintenance_interval_s = 60;

/// Bounds one blocklist download (`Manager.total_budget`). Generous, because a
/// megabyte-scale list over a household uplink is normal, and a download that
/// takes longer than this is not going to finish at all.
const download_budget_s = 300;

/// Per DoH upstream. `min_request_buf` is 512; the extra room costs nothing and
/// keeps a maximum-length name with a large OPT record comfortable.
const doh_request_buf_len = 1024;
const doh_transfer_buf_len = 4096;

pub fn run(runner: cli.Runner, args: cli.RunArgs) u8 {
    const code = serve(runner, args) catch |err| code: {
        runner.err.print("nxdns run failed: {s}\n", .{@errorName(err)}) catch {};
        const mapped = failureExitCode(err);
        if (mapped == cli.exit_check) {
            runner.err.writeAll("run `nxdns check` to see the configuration in full\n") catch {};
        }
        break :code mapped;
    };

    // Output the operator never received is not output, so a failed flush
    // outranks a successful run — the same rule `cli.finish` applies.
    runner.out.flush() catch return cli.exit_runtime;
    runner.err.flush() catch return cli.exit_runtime;
    return code;
}

/// The one classification, from `config/faults.zig`. This file keeps no list of
/// its own: `run` exiting 1 on a seed file `check` and `import` exit 2 on was
/// exactly the cost of the second list that used to be here.
fn failureExitCode(err: anyerror) u8 {
    return if (faults.isConfigFault(err)) cli.exit_check else cli.exit_runtime;
}

/// First run only: the file seeds an empty database and is ignored forever
/// after. Its diagnostics are the operator's one chance to see what the file
/// said, so they are printed the way `check` and `import` print them.
///
/// Printed on the way out either way. A seed file can be accepted and still
/// carry warnings — a blocklist source in no group is downloaded and compiled
/// into nothing — and a warning that only appears when the start fails is a
/// warning nobody ever reads: the start it describes is the one that worked.
/// `check` reported it and `run` did not, which left the same file graded two
/// ways.
///
/// The runner's error writer, not `std.log`: this runs before
/// `logging.install`, and one rendering of a diagnostic across `run`, `check`
/// and `import` is the point of `Diagnostics.writeAll`.
///
/// Flushed here rather than left to `run`'s exit flush. That writer is buffered
/// (`main` gives it 4 KiB) and `serve` does not return for as long as the
/// service runs, so a line left in the buffer reaches the operator when the
/// process stops — days after the start it describes. A failure path flushes
/// anyway because it returns immediately; the successful start is the one that
/// needs this.
fn seedFromFile(
    r: cli.Runner,
    config_db: *db.Db,
    dir: std.Io.Dir,
    config_path: []const u8,
) bootstrap.Error!bootstrap.Outcome {
    var diags: validate.Diagnostics = .init(r.gpa);
    defer diags.deinit();

    const result = bootstrap.bootstrap(r.io, r.gpa, config_db, dir, config_path, &diags);

    // Neither discard is an oversight, and the two answer different questions.
    //
    // On a rejected seed file, `result` is returned untouched: the operator gets
    // the reason the start failed, never a writer error standing in front of it.
    // A broken stderr is not why the configuration was refused.
    //
    // On a seed that worked, a failure here does not stop the start. The trade
    // is one lost warning line against a household with no name resolution, and
    // `run` before this point is the only stretch of this program where an
    // output failure could take DNS down at all — ruling 4 already says nothing
    // after it is fatal. Nor could the failure be reported: this writer *is* the
    // error channel, and `logging.install` has not run yet, so `std.log` resolves
    // to the same stderr a diagnostic about it would have to travel down.
    //
    // It is not lost from the process either. A failed drain consumes nothing,
    // so whatever the buffer held it still holds — that half is observed, in
    // "a broken error writer does not stop a first start that succeeded" below,
    // which reads the retained warning back out of the same writer.
    //
    // What happens to those bytes afterwards is derived, not watched, and is
    // labelled so deliberately. `Io.Writer.defaultFlush` drains while `end != 0`
    // and `run`'s exit flush maps a failure to exit 1, so a stderr still broken
    // at shutdown should carry the condition out in the exit code, and one that
    // recovered should deliver the line late. No test drives `run` that far.
    // Two limits come with the derivation: an empty buffer flushes clean and
    // reports nothing at all, and a failure that recovers ends at exit 0 with a
    // line the operator reads days after the start it describes.
    diags.writeAll(r.err) catch {};
    r.err.flush() catch {};
    return result;
}

fn serve(r: cli.Runner, args: cli.RunArgs) !u8 {
    const io = r.io;
    const gpa = r.gpa;
    const paths = args.paths;

    // -----------------------------------------------------------------------
    // storage and configuration
    // -----------------------------------------------------------------------

    var data = try cli.DataDir.open(io, gpa, paths.data_dir, true);
    defer data.close(io, gpa);

    var config_db = try data.openConfigDb(io);
    defer config_db.close();
    _ = try migrations.migrate(&config_db);

    _ = try seedFromFile(r, &config_db, std.Io.Dir.cwd(), paths.config);

    // Every string in `cfg` points into this arena, and the pool's endpoints,
    // the handler's records and the monitor's paths all keep such strings. It
    // therefore outlives all of them, which is why it is declared here and not
    // inside a narrower scope.
    var arena_state: std.heap.ArenaAllocator = .init(gpa);
    defer arena_state.deinit();
    const arena = arena_state.allocator();

    const cfg = try config_export.readConfig(&config_db, arena);

    // From here on `std.log` goes wherever the operator asked. Before this call
    // the sink passes through to stderr, which is where a failure above needs
    // to appear anyway.
    logging.install(io, cfg.logging);
    defer logging.deinstall();

    if (cfg.dns.rate_limit == 0 or cfg.dns.rate_window_seconds == 0) return error.BadRateLimit;

    // The API limiter and the session store assert these are nonzero
    // (`validate` refuses such a config, but nothing validates a database an
    // operator edited by hand), and a fault the operator can fix must exit 2,
    // not trip an assertion.
    if (cfg.web.enabled and
        (cfg.web.api_rate_limit_per_min == 0 or cfg.web.session_ttl_hours == 0))
    {
        return error.BadRateLimit;
    }

    // -----------------------------------------------------------------------
    // local answers
    // -----------------------------------------------------------------------

    // Ruling 12: the tables are published through the holder the API swaps, so
    // the holder owns them from here on and frees whichever generation is
    // current at shutdown.
    var tables: local_tables.LocalTables = .empty;
    defer tables.deinit(gpa);
    tables.records = try local_records.Records.build(gpa, cfg.local_records);
    tables.zones = try forward_zones.Zones.build(gpa, cfg.forward_zones);

    // -----------------------------------------------------------------------
    // blocklists
    // -----------------------------------------------------------------------

    // Two HTTP clients on purpose. A blocklist download streams tens of
    // megabytes and holds its connection for the whole of it; DoH queries must
    // not queue behind that, and the two have nothing to share but a type.
    var fetch_http: std.http.Client = .{ .allocator = gpa, .io = io };
    defer fetch_http.deinit();
    var dns_http: std.http.Client = .{ .allocator = gpa, .io = io };
    defer dns_http.deinit();

    const fetch_transfer = try gpa.alloc(u8, fetcher.min_transfer_buf);
    defer gpa.free(fetch_transfer);
    const fetch_redirect = try gpa.alloc(u8, fetcher.redirect_buffer_len);
    defer gpa.free(fetch_redirect);
    var fetch: fetcher.Fetcher = .{
        .http = &fetch_http,
        .transfer_buf = fetch_transfer,
        .redirect_buf = fetch_redirect,
    };

    var manager: manager_mod.Manager = try .init(
        gpa,
        &config_db,
        .{ .dir = data.dir },
        &fetch,
        cfg.blocklist_update,
        .{ .raw = .fromSeconds(download_budget_s), .clock = .awake },
    );
    defer manager.deinit(io);

    // -----------------------------------------------------------------------
    // upstream pool
    // -----------------------------------------------------------------------

    // Shared by every DoT client: the trust store does not vary per upstream
    // and the scan is expensive. Both must outlive the clients that hold them.
    var bundle: Certificate.Bundle = .empty;
    defer bundle.deinit(gpa);
    var bundle_lock: std.Io.RwLock = .init;

    var upstreams = try Upstreams.build(gpa, cfg.upstreams, &dns_http, &bundle, &bundle_lock);
    defer upstreams.deinit(gpa);

    var pool: pool_mod.Pool = .init(
        upstreams.active(),
        .{},
        .{ .raw = model.totalTimeout(cfg.upstream), .clock = .awake },
        @truncate(@as(u96, @bitCast(std.Io.Clock.real.now(io).nanoseconds))),
    );

    // -----------------------------------------------------------------------
    // per-query state
    // -----------------------------------------------------------------------

    var cache: dns_cache.DnsCache = try .init(gpa, cfg.cache);
    defer cache.deinit();

    var limiter: rate_limiter.RateLimiter = try .init(gpa, .{
        .limit = cfg.dns.rate_limit,
        .window_seconds = cfg.dns.rate_window_seconds,
    });
    defer limiter.deinit();

    var paused: pause.Pause = .{};
    var tracker: clients.Tracker = .init(cfg.logging.retention_days);

    // The queue holds waiting tasks in intrusive lists, so neither the buffer
    // nor the `Logger` may move once a task has touched either.
    const queue_buf = try gpa.alloc(logger_mod.Entry, cfg.logging.query_log_buffer_max);
    defer gpa.free(queue_buf);
    var query_logger: logger_mod.Logger = .init(cfg.logging, queue_buf);

    // Milestone 8 fans every logged query out to the SSE hub as well. The hub
    // exists only when the web interface does (ruling 6) — without it the sink
    // costs the query path one null check. Its rings are ~900 KiB, so it lives
    // on the heap and initializes in place; a by-value init would copy the
    // whole of it through this frame.
    var hub: ?*sse.Hub = null;
    defer if (hub) |hub_ptr| gpa.destroy(hub_ptr);
    if (cfg.web.enabled) {
        const created = try gpa.create(sse.Hub);
        created.init();
        hub = created;
    }
    var sink: query_sink.QuerySink = .init(&query_logger, hub);

    // -----------------------------------------------------------------------
    // disk, retention and the remaining connections (ruling 21)
    // -----------------------------------------------------------------------

    const data_path = try arena.dupeZ(u8, paths.data_dir);
    const log_dir_path: ?[:0]const u8 = if (cfg.logging.output == .file)
        try arena.dupeZ(u8, std.fs.path.dirname(cfg.logging.file_path) orelse ".")
    else
        null;
    var monitor: disk_monitor.Monitor = .init(cfg.disk, data.dir, data_path, log_dir_path);

    // Ruling 17. The scheduler consults it before every scheduled pass; the
    // startup `reload` below is an operator action and stays ungated.
    manager.monitor = &monitor;

    // One synchronous sample before anything can consult the gate. `Monitor`
    // initializes to `.ok`, and `Monitor.run` takes its first sample inside the
    // task — so without this call the log writer, the tracker and the blocklist
    // scheduler would all see "writes allowed" during the boot window and could
    // write to a critically full disk before the gate ever reflected it.
    // `run`'s own first sample repeats this one, which costs one extra statvfs
    // and directory scan at startup and keeps `run`'s "sample first, then
    // sleep" contract intact.
    //
    // `sample` returns nothing: every failure warns and counts inside the
    // monitor, and leaves the previous reading in place. A first sample that
    // fails therefore leaves the state at `.ok` — an unreadable filesystem is
    // not evidence that the disk is full, which is the monitor's documented
    // policy and the right one here too.
    monitor.sample(io);

    var retention: retention_mod.Retention = .init(cfg.logging);

    var querylog_writer_db = try data.openQuerylogDb(io);
    defer querylog_writer_db.close();
    var querylog_retention_db = try data.reopenQuerylogDb(io);
    defer querylog_retention_db.close();
    var tracker_db = try data.openConfigDb(io);
    defer tracker_db.close();

    // The web task's own two connections (ruling 26; m7 ruling 21: one SQLite
    // connection per task), opened only when the web interface is (ruling 6).
    // `reopenQuerylogDb` requires the file `openQuerylogDb` established above.
    var web_config_db: ?db.Db = null;
    defer if (web_config_db) |*database| database.close();
    var web_querylog_db: ?db.Db = null;
    defer if (web_querylog_db) |*database| database.close();
    if (cfg.web.enabled) {
        web_config_db = try data.openConfigDb(io);
        web_querylog_db = try data.reopenQuerylogDb(io);
    }

    var sessions: ?auth.Sessions = if (cfg.web.enabled) .init(cfg.web.session_ttl_hours) else null;

    var web_limiter: ?api_limiter.ApiLimiter = null;
    defer if (web_limiter) |*limiter_ptr| limiter_ptr.deinit();
    if (cfg.web.enabled) {
        web_limiter = try api_limiter.ApiLimiter.init(gpa, .{
            .rate_per_min = cfg.web.api_rate_limit_per_min,
            .localhost_exempt = cfg.web.api_localhost_exempt,
            .sse_max_per_ip = cfg.web.sse_max_connections_per_ip,
        });
    }

    // -----------------------------------------------------------------------
    // first snapshot
    // -----------------------------------------------------------------------

    // Ruling 4: serving starts either way. A household loses more from a DNS
    // server that refuses to start than from one that answers unfiltered until
    // the scheduler's first pass succeeds.
    manager.reload(io) catch |err| {
        log.warn(
            "loading the blocklist snapshot failed ({s}); serving unfiltered until the next refresh",
            .{@errorName(err)},
        );
    };

    // -----------------------------------------------------------------------
    // listeners
    // -----------------------------------------------------------------------

    var h: handler.Handler = .{
        .upstream = pool.client(),
        .blocking = .{ .mode = cfg.blocking.response, .ttl = cfg.blocking.ttl },
        .ecs_mode = cfg.edns.ecs_mode,
        .forward_read_timeout = .{ .raw = model.readTimeout(cfg.upstream), .clock = .awake },
        .manager = &manager,
        .local_tables = &tables,
        .cache = &cache,
        .negative_ttl_max = cfg.cache.negative_ttl_max,
        .limiter = &limiter,
        .sink = &sink,
        .pause = &paused,
        .tracker = &tracker,
    };

    // -----------------------------------------------------------------------
    // DoH/DoT listeners (milestone-10 ruling 11)
    // -----------------------------------------------------------------------

    // The stores are declared before the listeners on purpose: their deinit
    // defers run last, and `CertStore.deinit` asserts every connection has
    // released its generation, which only holds once the listeners are gone.
    // A certificate that does not load at boot is exit 2 — `nxdns check`
    // promises that an enabled endpoint has readable certs — while anything
    // that breaks later is the watcher's to absorb.
    var doh_certs: ?cert_store.CertStore = null;
    defer if (doh_certs) |*store| store.deinit(io);
    if (cfg.doh_server.enabled) {
        doh_certs = try openCertStore(r, gpa, io, cfg.doh_server, "doh_server", doh_server.alpn_protocols);
    }

    var dot_certs: ?cert_store.CertStore = null;
    defer if (dot_certs) |*store| store.deinit(io);
    if (cfg.dot_server.enabled) {
        dot_certs = try openCertStore(r, gpa, io, cfg.dot_server, "dot_server", dot_alpn);
    }

    // Bound here, in this frame, rather than through doh_server's module-level
    // entry: `/metrics` reads the listeners' counters through `WebState`, and
    // only a listener that lives in this frame has an address to wire there.
    // A failed bind warns and stays off (ruling 1, the web precedent): TLS DNS
    // failing to come up must not stop the plain-DNS side this box exists for.
    var doh: ?doh_server.DohServer = null;
    defer if (doh) |*server| server.deinit(gpa, io);
    if (doh_certs) |*store| doh = bindDoh(gpa, io, cfg.doh_server, &h, store);

    var dot: ?dot_server.DotServer = null;
    defer if (dot) |*server| server.deinit(io);
    if (dot_certs) |*store| dot = bindDot(gpa, io, cfg.dot_server, &h, store);

    // -----------------------------------------------------------------------
    // web interface (ruling 26)
    // -----------------------------------------------------------------------

    // Everything the web layer borrows lives above; the group below cancels the
    // web task before any of it is released. With the web interface disabled
    // the state stays in its null-defaulted shape and no task reads it.
    if (args.web_dev) |dir| web_dev_dir = dir;

    var web_state: web_server.WebState = .{ .gpa = gpa };
    // The live hash may own a gpa replacement after a settings PUT; this defer
    // runs after `group.cancel` below, so no web task can still read it.
    defer web_state.live_hash.deinit(gpa);
    if (cfg.web.enabled) web_state = .{
        .gpa = gpa,
        .web = cfg.web,
        .live_hash = .init(cfg.web.password_hash),
        .handler = &h,
        .pause = &paused,
        .tracker = &tracker,
        .manager = &manager,
        .pool = &pool,
        .monitor = &monitor,
        .local_tables = &tables,
        .logger = &query_logger,
        .retention = &retention,
        .sessions = if (sessions) |*s| s else null,
        .limiter = if (web_limiter) |*l| l else null,
        .hub = hub,
        .sink = &sink,
        .doh_certs = if (doh_certs) |*store| store else null,
        .dot_certs = if (dot_certs) |*store| store else null,
        .doh_listener = if (doh) |*server| server else null,
        .dot_listener = if (dot) |*server| server else null,
        .config_db = if (web_config_db) |*database| database else null,
        .querylog_db = if (web_querylog_db) |*database| database else null,
        .version = version.string,
        .started_unix = std.Io.Clock.real.now(io).toSeconds(),
        // Ruling 24: `--web-dev` serves from disk with no cache headers;
        // otherwise the embedded assets answer every non-/api miss.
        .fallback = if (args.web_dev != null) serveWebDev else static.fallback,
        .reload_fn = reloadManager,
    };

    const v6_bind = parseBind(r, cfg.dns.bind_ipv6, cfg.dns.port, "dns.bind_ipv6", .ip6) catch |err| return err;
    const v4_bind = parseBind(r, cfg.dns.bind_ipv4, cfg.dns.port, "dns.bind_ipv4", .ip4) catch |err| return err;

    // IPv6 first, and the order is load-bearing — see `Listeners`.
    var udp6: ?udp_server.UdpServer = udp_server.UdpServer.bind(gpa, io, v6_bind, &h, .{}) catch |err| bound: {
        if (!ipv6Unavailable(err)) return reportBind(r, "udp", v6_bind, err);
        log.warn("this system has no IPv6; serving IPv4 only", .{});
        break :bound null;
    };
    defer if (udp6) |*s| s.deinit(gpa, io);

    const v6_is_wildcard = udp6 != null and isWildcard(v6_bind);

    var udp4: ?udp_server.UdpServer = udp_server.UdpServer.bind(gpa, io, v4_bind, &h, .{}) catch |err| bound: {
        if (err != error.AddressInUse or !v6_is_wildcard) return reportBind(r, "udp", v4_bind, err);
        log.info("the IPv6 UDP listener is dual-stack and already serves IPv4", .{});
        break :bound null;
    };
    defer if (udp4) |*s| s.deinit(gpa, io);

    var tcp6: ?tcp_server.TcpServer = tcp_server.TcpServer.listen(gpa, io, v6_bind, &h, .{}) catch |err| bound: {
        if (!ipv6Unavailable(err)) return reportBind(r, "tcp", v6_bind, err);
        break :bound null;
    };
    defer if (tcp6) |*s| s.deinit(gpa, io);

    var tcp4: ?tcp_server.TcpServer = tcp_server.TcpServer.listen(gpa, io, v4_bind, &h, .{}) catch |err| bound: {
        if (err != error.AddressInUse or !(tcp6 != null and isWildcard(v6_bind))) {
            return reportBind(r, "tcp", v4_bind, err);
        }
        log.info("the IPv6 TCP listener is dual-stack and already serves IPv4", .{});
        break :bound null;
    };
    defer if (tcp4) |*s| s.deinit(gpa, io);

    // -----------------------------------------------------------------------
    // run
    // -----------------------------------------------------------------------

    shutdown.install(io);

    // Declared after everything it borrows, so its `cancel` — which both
    // requests cancellation and joins — is the first thing that runs on the way
    // out (ruling 22). Nothing below this line may be released while a task
    // could still touch it.
    var group: std.Io.Group = .init;
    defer group.cancel(io);

    if (udp6) |*s| try group.concurrent(io, udp_server.UdpServer.serve, .{ s, io });
    if (udp4) |*s| try group.concurrent(io, udp_server.UdpServer.serve, .{ s, io });
    if (tcp6) |*s| try group.concurrent(io, tcp_server.TcpServer.serve, .{ s, io });
    if (tcp4) |*s| try group.concurrent(io, tcp_server.TcpServer.serve, .{ s, io });
    if (doh) |*s| try group.concurrent(io, doh_server.DohServer.serve, .{ s, io });
    if (dot) |*s| try group.concurrent(io, dot_server.DotServer.serve, .{ s, io });
    if (doh_certs) |*store| try group.concurrent(io, cert_store.CertStore.watch, .{ store, io });
    if (dot_certs) |*store| try group.concurrent(io, cert_store.CertStore.watch, .{ store, io });

    const gate: ?*disk_monitor.Monitor = &monitor;
    try group.concurrent(io, logger_mod.Logger.runWriter, .{ &query_logger, io, &querylog_writer_db, gate });
    try group.concurrent(io, retention_mod.Retention.run, .{ &retention, io, &querylog_retention_db });
    try group.concurrent(io, disk_monitor.Monitor.run, .{ &monitor, io });
    try group.concurrent(io, manager_mod.Manager.runScheduler, .{ &manager, io });
    try group.concurrent(io, clients.Tracker.run, .{ &tracker, io, &tracker_db, gate });
    try group.concurrent(io, runMaintenance, .{ &h, io });

    // Started last (ruling 26), canceled by the same `group.cancel`; its inner
    // connection group is canceled, not awaited (ruling 4), so an idle
    // keep-alive client cannot hold shutdown open. A web bind failure is not
    // fatal: `web_server.serve` warns and returns, and the DNS side — the thing
    // this box exists for — keeps serving.
    if (cfg.web.enabled) try group.concurrent(io, web_server.serve, .{ &web_state, io });

    logStartup(io, &manager, upstreams.active().len, .{
        .udp6 = if (udp6) |*s| s.boundAddress() else null,
        .udp4 = if (udp4) |*s| s.boundAddress() else null,
        .tcp6 = if (tcp6) |*s| s.boundAddress() else null,
        .tcp4 = if (tcp4) |*s| s.boundAddress() else null,
    });

    // A canceled wait is a shutdown request too: whoever canceled this task
    // wants the process to stop, and the teardown below is how it stops.
    shutdown.wait(io) catch {};
    log.info("shutting down", .{});

    // Before the group is canceled, so the writer sees a closed queue and
    // drains what it holds rather than losing it to cancellation (ruling 22).
    query_logger.shutdown(io);

    return cli.exit_ok;
}

// ---------------------------------------------------------------------------
// web seams
// ---------------------------------------------------------------------------

/// Ruling 12: mutations to rules, blocklists, groups, clients and prefixes
/// rebuild the blocklist snapshot so the change is live on the next query. A
/// state without a manager has nothing to rebuild.
fn reloadManager(state: *web_server.WebState, io: std.Io) anyerror!void {
    const manager = state.manager orelse return;
    try manager.reload(io);
}

/// The `--web-dev` directory. `WebState.fallback` is a bare function pointer
/// with no closure to carry the path, and one process runs one composition
/// root, so the directory lives here: written once by `serve` before the web
/// task starts, read only by `serveWebDev`.
var web_dev_dir: []const u8 = "";

/// Dev-mode asset serving (ruling 24): straight from disk, no cache headers,
/// so an edit shows up on the next reload.
fn serveWebDev(
    state: *web_server.WebState,
    io: std.Io,
    request: *http_util.Request,
) http_util.HandlerError!void {
    _ = state;
    return static.serveFromDisk(web_dev_dir, io, request);
}

// ---------------------------------------------------------------------------
// DoH/DoT listeners
// ---------------------------------------------------------------------------

/// RFC 7858 has no IANA-registered ALPN id in wide use beyond "dot"
/// (milestone-10 ruling 5). Mbed TLS records the pointer, so the list must
/// outlive every `ServerContext` built with it; module scope gives it static
/// lifetime, the same shape as `doh_server.alpn_protocols`.
const dot_alpn: [*:null]const ?[*:0]const u8 = &.{"dot"};

/// The boot-time certificate load for one enabled endpoint. A failure is a
/// configuration fault the operator can fix — the same files `nxdns check`
/// verifies — reported in `check`'s style and mapped to exit 2 (ruling 11).
/// Out of memory is the one exception: nothing about the configuration is
/// wrong, so it keeps its own name and exits 1.
fn openCertStore(
    r: cli.Runner,
    gpa: Allocator,
    io: std.Io,
    endpoint: model.TlsEndpoint,
    section: []const u8,
    alpn: ?[*:null]const ?[*:0]const u8,
) !cert_store.CertStore {
    return cert_store.CertStore.init(gpa, io, endpoint.cert_path, endpoint.key_path, alpn) catch |err| {
        if (err == error.OutOfMemory) return error.OutOfMemory;
        r.err.print("{s}: '{s}' + '{s}': {s}\n", .{
            section,
            endpoint.cert_path,
            endpoint.key_path,
            cert_store.humanMessage(err),
        }) catch {};
        return error.BadCertificate;
    };
}

/// Ruling 1: a listener that cannot bind warns and stays off. The bind text
/// itself gets the same treatment — `validate` refuses it, but a hand-edited
/// database can still carry one, and it is not worth taking DNS down over.
fn bindDoh(
    gpa: Allocator,
    io: std.Io,
    endpoint: model.TlsEndpoint,
    h: *handler.Handler,
    store: *cert_store.CertStore,
) ?doh_server.DohServer {
    const bind_address = net.IpAddress.parse(endpoint.bind, endpoint.port) catch {
        log.warn("doh_server.bind '{s}' is not an IP address; DoH is disabled", .{endpoint.bind});
        return null;
    };
    const server = doh_server.DohServer.listen(gpa, io, bind_address, h, store, .{}) catch |err| {
        log.warn("doh listener cannot listen on {s}:{d}: {t}", .{ endpoint.bind, endpoint.port, err });
        return null;
    };
    log.info("doh listener on {f}", .{server.boundAddress()});
    return server;
}

fn bindDot(
    gpa: Allocator,
    io: std.Io,
    endpoint: model.TlsEndpoint,
    h: *handler.Handler,
    store: *cert_store.CertStore,
) ?dot_server.DotServer {
    const bind_address = net.IpAddress.parse(endpoint.bind, endpoint.port) catch {
        log.warn("dot_server.bind '{s}' is not an IP address; DoT is disabled", .{endpoint.bind});
        return null;
    };
    const server = dot_server.DotServer.listen(gpa, io, bind_address, h, store, .{}) catch |err| {
        log.warn("dot listener cannot listen on {s}:{d}: {t}", .{ endpoint.bind, endpoint.port, err });
        return null;
    };
    log.info("dot listener on {f}", .{server.boundAddress()});
    return server;
}

// ---------------------------------------------------------------------------
// background maintenance
// ---------------------------------------------------------------------------

/// Sweeps the cache and the rate-limiter table. Both are guarded by mutexes the
/// handler owns, because the handler is what contends for them; the sweeps live
/// here because walking a whole table is not work a query should pay for.
///
/// The locks are taken cancelably: unlike `handle`, this loop has an error
/// union to carry `error.Canceled` out of, and a shutdown that arrives while
/// the query path holds a lock should not wait for it.
fn runMaintenance(h: *handler.Handler, io: std.Io) std.Io.Cancelable!void {
    const interval: std.Io.Clock.Duration = .{
        .raw = .fromSeconds(maintenance_interval_s),
        .clock = .boot,
    };
    while (true) {
        try interval.sleep(io);

        if (h.cache) |cache| {
            const now_s = std.Io.Clock.real.now(io).toSeconds();
            try h.cache_mutex.lock(io);
            _ = cache.sweep(now_s);
            h.cache_mutex.unlock(io);
        }

        if (h.limiter) |limiter| {
            const now = std.Io.Clock.awake.now(io);
            try h.limiter_mutex.lock(io);
            _ = limiter.sweep(now);
            h.limiter_mutex.unlock(io);
        }
    }
}

// ---------------------------------------------------------------------------
// upstreams
// ---------------------------------------------------------------------------

/// The pool's entries and everything they point into.
///
/// `Pool.Entry.client` is a type-erased pointer into `doh` or `dot`, and each
/// client borrows a slice of `doh_buf`/`dot_buf`, so all five allocations live
/// exactly as long as the pool does. One entry is used by one task at a time
/// (`Entry.busy`), which is why the buffers are per client and not shared the
/// way `cli.probeUpstreams` shares them.
const Upstreams = struct {
    entries: []pool_mod.Entry,
    used: usize,
    doh: []doh_client.DohClient,
    dot: []dot_client.DotClient,
    doh_buf: []u8,
    dot_buf: []u8,

    /// A disabled upstream is left out entirely; a malformed one warns and is
    /// skipped, because one bad row in a table of four must not take DNS down.
    /// No usable row at all is a configuration fault.
    fn build(
        gpa: Allocator,
        servers: []const model.UpstreamServer,
        http: *std.http.Client,
        bundle: *Certificate.Bundle,
        bundle_lock: *std.Io.RwLock,
    ) (Allocator.Error || error{NoUsableUpstreams})!Upstreams {
        var enabled: usize = 0;
        for (servers) |server| {
            if (server.enabled) enabled += 1;
        }
        if (enabled == 0) return error.NoUsableUpstreams;

        const chunk = tls.Client.min_buffer_len;

        var self: Upstreams = .{
            .entries = try gpa.alloc(pool_mod.Entry, enabled),
            .used = 0,
            .doh = &.{},
            .dot = &.{},
            .doh_buf = &.{},
            .dot_buf = &.{},
        };
        errdefer self.deinit(gpa);

        self.doh = try gpa.alloc(doh_client.DohClient, enabled);
        self.dot = try gpa.alloc(dot_client.DotClient, enabled);
        self.doh_buf = try gpa.alloc(u8, enabled * (doh_request_buf_len + doh_transfer_buf_len));
        self.dot_buf = try gpa.alloc(u8, enabled * 4 * chunk);

        var doh_count: usize = 0;
        var dot_count: usize = 0;

        for (servers) |server| {
            if (!server.enabled) continue;

            const endpoint = transport.Endpoint.parse(server.url) catch {
                log.warn(
                    "upstream {f} is not an https:// or tls:// endpoint; skipped",
                    .{safe_url.redactQuoted(server.url)},
                );
                continue;
            };

            const client: transport.Client = switch (endpoint.scheme) {
                .doh => doh: {
                    const base = doh_count * (doh_request_buf_len + doh_transfer_buf_len);
                    const slot = &self.doh[doh_count];
                    slot.* = doh_client.DohClient.init(
                        http,
                        endpoint,
                        self.doh_buf[base..][0..doh_request_buf_len],
                        self.doh_buf[base + doh_request_buf_len ..][0..doh_transfer_buf_len],
                    ) catch {
                        log.warn(
                            "upstream {f} is not a usable DoH url; skipped",
                            .{safe_url.redactQuoted(server.url)},
                        );
                        continue;
                    };
                    doh_count += 1;
                    break :doh slot.client();
                },
                .dot => dot: {
                    const base = dot_count * 4 * chunk;
                    const slot = &self.dot[dot_count];
                    slot.* = dot_client.DotClient.init(endpoint, server.tls_name, gpa, bundle, bundle_lock, .{
                        .tls_read = self.dot_buf[base..][0..chunk],
                        .tls_write = self.dot_buf[base + chunk ..][0..chunk],
                        .stream_read = self.dot_buf[base + 2 * chunk ..][0..chunk],
                        .stream_write = self.dot_buf[base + 3 * chunk ..][0..chunk],
                    });
                    dot_count += 1;
                    break :dot slot.client();
                },
            };

            self.entries[self.used] = .{
                .endpoint = endpoint,
                .client = client,
                .priority = server.priority,
                .enabled = true,
                .health = .init,
            };
            self.used += 1;
        }

        if (self.used == 0) return error.NoUsableUpstreams;
        return self;
    }

    /// The prefix `Pool.init` is given. The rest of `entries` is allocated but
    /// never filled, which is what keeps `deinit` able to free the whole block.
    fn active(self: *Upstreams) []pool_mod.Entry {
        return self.entries[0..self.used];
    }

    fn deinit(self: *Upstreams, gpa: Allocator) void {
        gpa.free(self.dot_buf);
        gpa.free(self.doh_buf);
        gpa.free(self.dot);
        gpa.free(self.doh);
        gpa.free(self.entries);
        self.* = undefined;
    }
};

// ---------------------------------------------------------------------------
// listeners
// ---------------------------------------------------------------------------

/// The addresses actually bound, for the startup line. A null is a family this
/// process does not listen on separately — either because the system has no
/// IPv6, or because the IPv6 socket is dual-stack and already serves IPv4.
const Listeners = struct {
    udp6: ?net.IpAddress,
    udp4: ?net.IpAddress,
    tcp6: ?net.IpAddress,
    tcp4: ?net.IpAddress,
};

const BindFamily = enum { ip4, ip6 };

/// `config/validate.checkBind` enforces the same family rule on import, check
/// and settings PUT — but not on a config.db written before the rule existed,
/// and `serve` loads that DB without re-validating. Boot is the last seam: a
/// cross-family literal here would bind the wrong family's socket and make the
/// real one fail with AddressInUse, silently losing a family.
fn parseBind(
    r: cli.Runner,
    text: []const u8,
    port: u16,
    field: []const u8,
    family: BindFamily,
) !net.IpAddress {
    const addr = net.IpAddress.parse(text, port) catch {
        r.err.print("{s}: '{s}' is not an IP address\n", .{ field, text }) catch {};
        return error.BadBindAddress;
    };
    const matches = switch (addr) {
        .ip4 => family == .ip4,
        .ip6 => family == .ip6,
    };
    if (!matches) {
        const digit: u8 = if (family == .ip4) '4' else '6';
        r.err.print(
            "{s}: '{s}' is not an IPv{c} address; re-import the configuration or correct it with a settings PUT\n",
            .{ field, text, digit },
        ) catch {};
        return error.BadBindAddress;
    }
    return addr;
}

test "parseBind refuses a bind address of the wrong family" {
    var out_buf: [8]u8 = undefined;
    var err_buf: [256]u8 = undefined;
    var out: Writer = .fixed(&out_buf);
    var err_writer: Writer = .fixed(&err_buf);
    const r: cli.Runner = .{
        .io = std.testing.io,
        .gpa = std.testing.allocator,
        .out = &out,
        .err = &err_writer,
    };

    _ = try parseBind(r, "0.0.0.0", 53, "dns.bind_ipv4", .ip4);
    _ = try parseBind(r, "::", 53, "dns.bind_ipv6", .ip6);

    try std.testing.expectError(
        error.BadBindAddress,
        parseBind(r, "0.0.0.0", 53, "dns.bind_ipv6", .ip6),
    );
    try std.testing.expectError(
        error.BadBindAddress,
        parseBind(r, "::", 53, "dns.bind_ipv4", .ip4),
    );

    const printed = err_writer.buffered();
    try std.testing.expect(std.mem.containsAtLeast(u8, printed, 1, "is not an IPv6 address"));
    try std.testing.expect(std.mem.containsAtLeast(u8, printed, 1, "is not an IPv4 address"));
}

test "run maps a rejected configuration to exit 2 and everything else to exit 1" {
    try std.testing.expectEqual(cli.exit_check, failureExitCode(error.MissingDefaultGroup));
    try std.testing.expectEqual(cli.exit_check, failureExitCode(error.ParseZon));
    try std.testing.expectEqual(cli.exit_check, failureExitCode(error.NoUsableUpstreams));
    try std.testing.expectEqual(cli.exit_check, failureExitCode(error.BadCertificate));
    try std.testing.expectEqual(cli.exit_runtime, failureExitCode(error.AccessDenied));
    try std.testing.expectEqual(cli.exit_runtime, failureExitCode(error.OutOfMemory));
}

test "run, check and import agree on a seed file with no default group" {
    const config_import = @import("config/import.zig");

    var threaded: std.Io.Threaded = .init(std.testing.allocator, .{});
    defer threaded.deinit();
    const io = threaded.io();
    const gpa = std.testing.allocator;

    // The file from discrepancy D1: parseable, one upstream, no group named
    // 'default'.
    const source: [:0]const u8 =
        \\.{
        \\    .groups = .{ .{ .name = "kids" } },
        \\    .upstreams = .{ .{ .url = "https://dns.example/dns-query" } },
        \\}
    ;

    // `run`: `serve` seeds through `bootstrap`, which is a wrapper over this
    // exact call, so this is the error `run` classifies.
    var database = try db.Db.open(":memory:", .{ .mode = .memory });
    defer database.close();
    try db.applyPragmas(&database, .{});
    _ = try migrations.migrate(&database);

    var diags: validate.Diagnostics = .init(gpa);
    defer diags.deinit();
    try std.testing.expectError(
        error.MissingDefaultGroup,
        config_import.importSource(io, gpa, &database, source, .{}, &diags),
    );
    try std.testing.expectEqual(cli.exit_check, failureExitCode(error.MissingDefaultGroup));

    // `import`: `cli.failureExitCode` reaches exit 2 by either route — the
    // recorded failures, or the classification `run` just used.
    try std.testing.expect(diags.failureCount() != 0);
    try std.testing.expect(faults.isConfigFault(error.MissingDefaultGroup));

    // `check`: the same file, through the code `nxdns check` runs.
    var arena_state: std.heap.ArenaAllocator = .init(gpa);
    defer arena_state.deinit();
    const cfg = try std.zon.parse.fromSliceAlloc(model.Config, arena_state.allocator(), source, null, .{});

    var out_buf: [2048]u8 = undefined;
    var err_buf: [256]u8 = undefined;
    var out: Writer = .fixed(&out_buf);
    var err_writer: Writer = .fixed(&err_buf);
    const r: cli.Runner = .{ .io = io, .gpa = gpa, .out = &out, .err = &err_writer };

    try std.testing.expectEqual(cli.exit_check, try cli.checkConfig(r, cfg, false));
}

test "a configuration whose blocklist source is in no group imports and checks clean" {
    const config_import = @import("config/import.zig");

    var threaded: std.Io.Threaded = .init(std.testing.allocator, .{});
    defer threaded.deinit();
    const io = threaded.io();
    const gpa = std.testing.allocator;

    // A source is created before it is attached — `docs/tutorial/first-run.md`
    // POSTs the blocklist and then PUTs the group's sources — so an unattached
    // source is a legal intermediate state on every write path. It warns; it
    // never fails.
    const source: [:0]const u8 =
        \\.{
        \\    .groups = .{ .{ .name = "default" } },
        \\    .upstreams = .{ .{ .url = "https://dns.example/dns-query" } },
        \\    .blocklist_sources = .{ .{ .url = "https://lists.example/hosts.txt", .name = "ads" } },
        \\}
    ;

    var database = try db.Db.open(":memory:", .{ .mode = .memory });
    defer database.close();
    try db.applyPragmas(&database, .{});
    _ = try migrations.migrate(&database);

    var diags: validate.Diagnostics = .init(gpa);
    defer diags.deinit();
    try config_import.importSource(io, gpa, &database, source, .{}, &diags);
    try std.testing.expectEqual(@as(usize, 0), diags.failureCount());
    try std.testing.expectEqual(@as(usize, 1), diags.warningCount());

    var arena_state: std.heap.ArenaAllocator = .init(gpa);
    defer arena_state.deinit();
    const cfg = try std.zon.parse.fromSliceAlloc(model.Config, arena_state.allocator(), source, null, .{});

    var check_diags: validate.Diagnostics = .init(gpa);
    defer check_diags.deinit();
    // No error means no subcommand may reject it; the warning is report-only.
    try validate.validate(cfg, &check_diags);
    try std.testing.expectEqual(@as(usize, 0), check_diags.failureCount());
    try std.testing.expectEqual(@as(usize, 1), check_diags.warningCount());
}

test "a first start that seeds from a file prints the warnings the file earned" {
    // D5, second half. The seed file is read once in the life of a database, so
    // a warning it earns is printed on that start or never. `run` printed
    // diagnostics only when the file was rejected, which made a successful first
    // start the one place the finding could not surface.
    var threaded: std.Io.Threaded = .init(std.testing.allocator, .{});
    defer threaded.deinit();
    const io = threaded.io();
    const gpa = std.testing.allocator;

    var tmp = std.testing.tmpDir(.{});
    defer tmp.cleanup();
    try tmp.dir.writeFile(io, .{ .sub_path = "config.zon", .data =
        \\.{
        \\    .groups = .{ .{ .name = "default" } },
        \\    .upstreams = .{ .{ .url = "https://dns.example/dns-query" } },
        \\    .blocklist_sources = .{ .{ .url = "https://lists.example/hosts.txt", .name = "ads" } },
        \\}
    });

    var database = try db.Db.open(":memory:", .{ .mode = .memory });
    defer database.close();
    try db.applyPragmas(&database, .{});
    _ = try migrations.migrate(&database);

    // A buffered file writer, the shape `main` builds over stderr, and not
    // `Writer.fixed`: a fixed writer's flush is a no-op, so it counts a line
    // still sitting in the buffer as delivered. Reading the file back is the
    // only way to ask what the operator can actually see.
    var out_buf: [64]u8 = undefined;
    var out: Writer = .fixed(&out_buf);
    var err_file = try tmp.dir.createFile(io, "stderr.txt", .{});
    defer err_file.close(io);
    var err_buf: [4096]u8 = undefined;
    var err_writer = err_file.writer(io, &err_buf);
    const r: cli.Runner = .{ .io = io, .gpa = gpa, .out = &out, .err = &err_writer.interface };

    // The file is valid, so the start succeeds and the database is seeded.
    try std.testing.expectEqual(
        bootstrap.Outcome.seeded,
        try seedFromFile(r, &database, tmp.dir, "config.zon"),
    );
    try std.testing.expectEqual(@as(i64, 1), try database.queryInt("SELECT count(*) FROM upstreams"));

    // Nothing flushes here on purpose. In production `serve` runs from this
    // point until the service stops, so a line that has not reached the file by
    // now is a line the operator does not get for days.
    const printed = try tmp.dir.readFileAlloc(io, "stderr.txt", gpa, .limited(8192));
    defer gpa.free(printed);

    try std.testing.expect(std.mem.containsAtLeast(u8, printed, 1, "WARN blocklist_sources[0]: "));
    try std.testing.expect(std.mem.containsAtLeast(u8, printed, 1, "belongs to no group"));
    // A warning is not a rejection: nothing here claims the start failed.
    try std.testing.expectEqual(@as(usize, 0), std.mem.count(u8, printed, "FAIL"));
}

/// A broken stderr, in the shape `main` builds: a buffered `File.Writer`, with
/// its drain switched to the mode that fails. `Writer.fixed` cannot stand in —
/// its flush is `noopFlush`, so it has no failure to report and its `written()`
/// counts a line still sitting in the buffer as delivered.
///
/// The file stays empty for as long as the mode is `.failure`, which is what
/// lets a test tell "the writer really failed" from "the writer worked".
fn brokenErrWriter(io: std.Io, file: std.Io.File, buffer: []u8) std.Io.File.Writer {
    var w = file.writer(io, buffer);
    w.mode = .failure;
    return w;
}

test "a broken error writer does not replace the reason a seed file was rejected" {
    // The operator has to see why seeding failed, and a broken stderr is not
    // that reason.
    var threaded: std.Io.Threaded = .init(std.testing.allocator, .{});
    defer threaded.deinit();
    const io = threaded.io();
    const gpa = std.testing.allocator;

    var tmp = std.testing.tmpDir(.{});
    defer tmp.cleanup();
    // No group named 'default': rejected, and it records a FAIL line on the way.
    try tmp.dir.writeFile(io, .{ .sub_path = "config.zon", .data =
        \\.{
        \\    .groups = .{ .{ .name = "kids" } },
        \\    .upstreams = .{ .{ .url = "https://dns.example/dns-query" } },
        \\}
    });

    var database = try db.Db.open(":memory:", .{ .mode = .memory });
    defer database.close();
    try db.applyPragmas(&database, .{});
    _ = try migrations.migrate(&database);

    var out_buf: [64]u8 = undefined;
    var out: Writer = .fixed(&out_buf);
    var err_file = try tmp.dir.createFile(io, "stderr.txt", .{});
    defer err_file.close(io);
    // Eight bytes: no FAIL line fits, so `writeAll` must drain mid-line and is
    // itself the call that fails. The test below covers the other discard, where
    // the line fits and only the flush fails.
    var err_buf: [8]u8 = undefined;
    var err_writer = brokenErrWriter(io, err_file, &err_buf);
    const r: cli.Runner = .{ .io = io, .gpa = gpa, .out = &out, .err = &err_writer.interface };

    try std.testing.expectError(
        error.MissingDefaultGroup,
        seedFromFile(r, &database, tmp.dir, "config.zon"),
    );

    // Empty, so the writer did fail — without this the assertion above would
    // hold just as well against a writer that worked.
    const printed = try tmp.dir.readFileAlloc(io, "stderr.txt", gpa, .limited(8192));
    defer gpa.free(printed);
    try std.testing.expectEqual(@as(usize, 0), printed.len);
}

test "a broken error writer does not stop a first start that succeeded" {
    // The call this file makes: a DNS server for a household does not refuse to
    // resolve because stderr is broken. What it must not do is drop the warning
    // on the floor, so the second half checks the buffer still holds it.
    var threaded: std.Io.Threaded = .init(std.testing.allocator, .{});
    defer threaded.deinit();
    const io = threaded.io();
    const gpa = std.testing.allocator;

    var tmp = std.testing.tmpDir(.{});
    defer tmp.cleanup();
    // Valid, and it earns one warning: the source belongs to no group.
    try tmp.dir.writeFile(io, .{ .sub_path = "config.zon", .data =
        \\.{
        \\    .groups = .{ .{ .name = "default" } },
        \\    .upstreams = .{ .{ .url = "https://dns.example/dns-query" } },
        \\    .blocklist_sources = .{ .{ .url = "https://lists.example/hosts.txt", .name = "ads" } },
        \\}
    });

    var database = try db.Db.open(":memory:", .{ .mode = .memory });
    defer database.close();
    try db.applyPragmas(&database, .{});
    _ = try migrations.migrate(&database);

    var out_buf: [64]u8 = undefined;
    var out: Writer = .fixed(&out_buf);
    var err_file = try tmp.dir.createFile(io, "stderr.txt", .{});
    defer err_file.close(io);
    // 4 KiB, the buffer `main` gives the real stderr writer: the warning fits,
    // so `writeAll` succeeds into the buffer and the flush is what fails. That
    // is the shape production hits.
    var err_buf: [4096]u8 = undefined;
    var err_writer = brokenErrWriter(io, err_file, &err_buf);
    const r: cli.Runner = .{ .io = io, .gpa = gpa, .out = &out, .err = &err_writer.interface };

    try std.testing.expectEqual(
        bootstrap.Outcome.seeded,
        try seedFromFile(r, &database, tmp.dir, "config.zon"),
    );
    try std.testing.expectEqual(@as(i64, 1), try database.queryInt("SELECT count(*) FROM upstreams"));

    // Nothing reached the file, so the flush really did fail.
    const undelivered = try tmp.dir.readFileAlloc(io, "stderr.txt", gpa, .limited(8192));
    defer gpa.free(undelivered);
    try std.testing.expectEqual(@as(usize, 0), undelivered.len);

    // And the warning is still buffered rather than dropped: this is the same
    // writer, and these are the bytes `run`'s exit flush meets on the way out.
    err_writer.mode = .positional;
    try err_writer.interface.flush();
    const printed = try tmp.dir.readFileAlloc(io, "stderr.txt", gpa, .limited(8192));
    defer gpa.free(printed);
    try std.testing.expect(std.mem.containsAtLeast(u8, printed, 1, "WARN blocklist_sources[0]: "));
}

fn reportBind(r: cli.Runner, which: []const u8, addr: net.IpAddress, err: anyerror) anyerror {
    r.err.print("cannot bind {s} {f}: {s}\n", .{ which, addr, @errorName(err) }) catch {};
    return err;
}

/// The kernel has no IPv6 at all. Refusing to start would be the wrong answer
/// for the default configuration, which asks for both families.
fn ipv6Unavailable(err: anyerror) bool {
    return switch (err) {
        error.AddressFamilyUnsupported,
        error.ProtocolUnsupportedBySystem,
        error.ProtocolUnsupportedByAddressFamily,
        => true,
        else => false,
    };
}

fn isWildcard(addr: net.IpAddress) bool {
    switch (addr) {
        .ip4 => |a| {
            for (a.bytes) |b| if (b != 0) return false;
        },
        .ip6 => |a| {
            for (a.bytes) |b| if (b != 0) return false;
        },
    }
    return true;
}

// ---------------------------------------------------------------------------
// startup line
// ---------------------------------------------------------------------------

/// One line, at info, naming what an operator needs to see in `journalctl`
/// right after a restart: where it listens, how many upstreams it has, and
/// whether filtering is live.
fn logStartup(io: std.Io, manager: *manager_mod.Manager, upstream_count: usize, bound: Listeners) void {
    var buf: [256]u8 = undefined;
    var w: Writer = .fixed(&buf);
    appendBind(&w, "udp", bound.udp6);
    appendBind(&w, "udp", bound.udp4);
    appendBind(&w, "tcp", bound.tcp6);
    appendBind(&w, "tcp", bound.tcp4);

    // The generation is read under the snapshot handle's shared lock, which is
    // the same lock the reload took to publish it.
    if (manager.acquire(io)) |snapshot| {
        defer snapshot.release(io);
        log.info("nxdns {s} serving on{s}; {d} upstream(s); blocklist generation {d}", .{
            version.string,
            w.buffered(),
            upstream_count,
            manager.generation,
        });
    } else {
        log.info("nxdns {s} serving on{s}; {d} upstream(s); unfiltered (no blocklist snapshot)", .{
            version.string,
            w.buffered(),
            upstream_count,
        });
    }
}

/// Silent on overflow: a truncated startup line is not worth a failure path,
/// and 256 bytes hold four addresses.
fn appendBind(w: *Writer, which: []const u8, addr: ?net.IpAddress) void {
    const value = addr orelse return;
    w.print(" {s} {f}", .{ which, value }) catch {};
}