resolver: try servers in a random order on each resolution (closes #138)

Every resolution walked the root servers in a fixed order, so
a.root-servers.net got every first query and its timeouts were paid on
every lookup. Each list of servers the resolver walks, the root servers
and the nameservers of each zone below them, is now walked in a random
order from the standard library's rand.Shuffle, chosen anew each time.
Failover is unchanged: after a failure the next server is tried, and
the walk fails only when every server has. The shuffle is passed in, so
the tests check the order with a seeded source; which server a live
query reached is not observable, so no test fails if the walk stops
shuffling.

Model: opus-5-5

README.md

@@ -110,9 +110,7 @@ Contributions that introduce mocked, faked, or stubbed DNS will be rejected.
     - Port transitioned from open to closed (or vice versa).
     - New IP appeared (from DNS change) and its port state was recorded.
     - IP disappeared (from DNS change) — noted in the DNS change notification;
-      port state for that IP is removed. When none of a name's nameservers
+      port state for that IP is removed.
-      answered, its addresses are not known, so the port state saved for them is
-      kept.
 
 ### TLS Certificate Monitoring
 
@@ -402,9 +400,8 @@ performs full iterative resolution:
    directly for the requested records.
 
 In steps 2 and 3 the servers are asked one at a time in a random order, chosen
-anew each time, so no one root server gets every first query. A server that does
+anew each time, until one replies; the lookup fails only when none does. No one
-not reply, or refuses the query, is passed over for the next one; the first
+root server gets every first query.
-other reply is used, even a SERVFAIL, and no further server is asked.
 
 This approach ensures:
 

TODO.md

@@ -21,8 +21,6 @@ trial run of the finished image: https://git.eeqj.de/sneak/dnswatcher/issues/149
 
 - 2026-10-01: the resolver tries root servers, and every other server list it
   walks, in a random order each time, not always from the top (closes #138).
-- 2026-10-01: when none of a configured name's nameservers answered, the port
-  state saved for its addresses is kept, not removed (closes #193).
 - 2026-10-01: `ResolveIPAddresses` returns an error, not no addresses, when no
   nameserver of the name's zone answered (closes #190).
 - 2026-10-01: `make fmt` and `make fmt-check` cover Markdown with prettier, run

internal/watcher/export_test.go

@@ -67,11 +67,6 @@ func (w *Watcher) DetectNSAddressChanges(
 	w.detectNSAddressChanges(ctx, domain, prev, current)
 }
 
-// CheckAllPorts exports checkAllPorts for testing.
-func (w *Watcher) CheckAllPorts(ctx context.Context) {
-	w.checkAllPorts(ctx)
-}
-
 // BuildHostnameState exports buildHostnameState for testing.
 func BuildHostnameState(
 	results map[string]*resolver.NameserverResponse,

internal/watcher/nsfailure_test.go

@@ -331,118 +331,3 @@ func TestNameserverThatRefuses(t *testing.T) {
 		)
 	}
 }
-
-// TestPortStateWhenNoNameserverAnswered runs the port checks on
-// hostname state built here, which gives the name no address. The port
-// state saved for its old address is kept only when the name is a
-// configured hostname or domain and none of its nameservers answered.
-func TestPortStateWhenNoNameserverAnswered(t *testing.T) {
-	t.Parallel()
-
-	noneAnswered := saved(map[string]*state.NameserverRecordState{
-		nsA: failed(), nsB: failed(),
-	})
-	oneAnsweredNoAddress := saved(map[string]*state.NameserverRecordState{
-		nsA: answered(map[string][]string{}), nsB: failed(),
-	})
-
-	configured := []string{host}
-
-	tests := []struct {
-		name      string
-		hostname  *state.HostnameState
-		hostnames []string
-		domains   []string
-		wantKept  bool
-	}{
-		{"no nameserver answered", noneAnswered, configured, nil, true},
-		{
-			"no nameserver answered, configured as a domain",
-			noneAnswered, nil, configured, true,
-		},
-		{
-			"one answered with no address",
-			oneAnsweredNoAddress, configured, nil, false,
-		},
-		{"no nameserver answered, not configured", noneAnswered, nil, nil, false},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			t.Parallel()
-
-			cfg := defaultTestConfig(t)
-			cfg.Hostnames = tt.hostnames
-			cfg.Domains = tt.domains
-
-			// The port checks read the saved hostname state and look
-			// nothing up, so the watcher has no resolver.
-			deps := newTestDeps(t, cfg)
-			w := watcher.NewForTest(
-				cfg, deps.state, nil,
-				deps.portChecker, deps.tlsChecker, deps.notifier,
-			)
-
-			key := ip1 + ":443"
-
-			deps.state.SetHostnameState(host, tt.hostname)
-			deps.state.SetPortState(key, &state.PortState{
-				Open: true, Hostnames: []string{host},
-			})
-
-			w.CheckAllPorts(t.Context())
-
-			_, kept := deps.state.GetPortState(key)
-			if kept != tt.wantKept {
-				t.Errorf("port state %s kept: %v, want %v", key, kept, tt.wantKept)
-			}
-		})
-	}
-}
-
-// TestPortStateWhenNoNameserverAnsweredAndOtherNameMovesAway saves the
-// port state of an address two configured hostnames resolve to. While
-// none of the first one's nameservers answer, the port checks run with
-// the other one still at that address, then after it moved away; the
-// port state is kept both times.
-func TestPortStateWhenNoNameserverAnsweredAndOtherNameMovesAway(
-	t *testing.T,
-) {
-	t.Parallel()
-
-	const other = "mail.example.net"
-
-	cfg := defaultTestConfig(t)
-	cfg.Hostnames = []string{host, other}
-
-	// The port checks read the saved hostname state and look nothing
-	// up, so the watcher has no resolver.
-	deps := newTestDeps(t, cfg)
-	w := watcher.NewForTest(
-		cfg, deps.state, nil,
-		deps.portChecker, deps.tlsChecker, deps.notifier,
-	)
-
-	key := ip1 + ":443"
-
-	deps.state.SetPortState(key, &state.PortState{
-		Open: true, Hostnames: []string{host, other},
-	})
-	deps.state.SetHostnameState(host, saved(
-		map[string]*state.NameserverRecordState{nsA: failed(), nsB: failed()},
-	))
-
-	for _, otherIP := range []string{ip1, ip2} {
-		deps.state.SetHostnameState(other, saved(
-			map[string]*state.NameserverRecordState{
-				nsA: answered(map[string][]string{"A": {otherIP}}),
-			},
-		))
-
-		w.CheckAllPorts(t.Context())
-
-		if _, kept := deps.state.GetPortState(key); !kept {
-			t.Fatalf("port state %s removed with %s at %s", key, other, otherIP)
-		}
-	}
-}

internal/watcher/watcher.go

@@ -4,7 +4,6 @@ import (
 	"context"
 	"fmt"
 	"log/slog"
-	"slices"
 	"sort"
 	"strings"
 	"sync"
@@ -709,46 +708,15 @@ func parsePortKey(key string) (string, int) {
 }
 
 // cleanupStalePorts removes port state entries that are no
-// longer referenced by any hostname in the current DNS data. An
+// longer referenced by any hostname in the current DNS data.
-// entry saved for a configured name none of whose nameservers
-// answered is kept: that name's addresses are not known, not gone.
 func (w *Watcher) cleanupStalePorts(
 	currentAssociations map[string][]string,
 ) {
 	for _, key := range w.state.GetAllPortKeys() {
-		if _, exists := currentAssociations[key]; exists {
+		if _, exists := currentAssociations[key]; !exists {
-			continue
+			w.state.DeletePortState(key)
-		}
-
-		ps, ok := w.state.GetPortState(key)
-		if ok && slices.ContainsFunc(ps.Hostnames, w.noNameserverAnswered) {
-			continue
-		}
-
-		w.state.DeletePortState(key)
-	}
-}
-
-// noNameserverAnswered reports whether name is a configured domain or
-// hostname and none of its nameservers answered on its last check.
-func (w *Watcher) noNameserverAnswered(name string) bool {
-	if !slices.Contains(w.config.Hostnames, name) &&
-		!slices.Contains(w.config.Domains, name) {
-		return false
-	}
-
-	hs, ok := w.state.GetHostnameState(name)
-	if !ok {
-		return false
-	}
-
-	for _, nsState := range hs.RecordsByNameserver {
-		if nsState.Status == statusOK {
-			return false
 		}
 	}
-
-	return true
 }
 
 func (w *Watcher) collectIPs(hostname string) []string {
@@ -827,24 +795,9 @@ func (w *Watcher) checkSinglePort(
 		)
 	}
 
-	// A configured name on the saved list none of whose nameservers
-	// answered stays on it, so the entry is kept when the other names
-	// stop resolving to this address.
-	savedHostnames := slices.Clone(hostnames)
-
-	if hasPrev {
-		for _, name := range prev.Hostnames {
-			if !slices.Contains(hostnames, name) && w.noNameserverAnswered(name) {
-				savedHostnames = append(savedHostnames, name)
-			}
-		}
-
-		sort.Strings(savedHostnames)
-	}
-
 	w.state.SetPortState(key, &state.PortState{
 		Open:        result.Open,
-		Hostnames:   savedHostnames,
+		Hostnames:   hostnames,
 		LastChecked: now,
 	})
 }