dnswatcher/internal/resolver/resolver_test.go

package resolver_test

import (
	"context"
	"fmt"
	"log/slog"
	"net"
	"os"
	"sort"
	"strings"
	"testing"
	"time"

	"github.com/stretchr/testify/assert"
	"github.com/stretchr/testify/require"

	"sneak.berlin/go/dnswatcher/internal/livednstest"
	"sneak.berlin/go/dnswatcher/internal/resolver"
)

// ----------------------------------------------------------------
// Test helpers
// ----------------------------------------------------------------

func newTestResolver(t *testing.T) *resolver.Resolver {
	t.Helper()

	log := slog.New(slog.NewTextHandler(
		os.Stderr,
		&slog.HandlerOptions{Level: slog.LevelDebug},
	))

	return resolver.NewFromLogger(log)
}

// findOneNSForDomain picks one authoritative nameserver to aim a
// test at. Quorum handling lives in livedns_test.go, and the live-DNS
// retry and concurrency limit in package livednstest.
func findOneNSForDomain(
	t *testing.T,
	r *resolver.Resolver,
	domain string,
) string {
	t.Helper()

	return liveFindAuthoritative(t, r, domain)[0]
}

// ----------------------------------------------------------------
// FindAuthoritativeNameservers tests
// ----------------------------------------------------------------

func TestFindAuthoritativeNameservers_ValidDomain(
	t *testing.T,
) {
	t.Parallel()

	r := newTestResolver(t)
	nameservers := liveFindAuthoritative(t, r, "google.com")

	hasGoogleNS := false

	for _, ns := range nameservers {
		if strings.Contains(ns, "google") {
			hasGoogleNS = true

			break
		}
	}

	assert.True(t, hasGoogleNS,
		"expected google nameservers, got: %v", nameservers,
	)
}

func TestFindAuthoritativeNameservers_Subdomain(
	t *testing.T,
) {
	t.Parallel()

	r := newTestResolver(t)
	fromHost := liveFindAuthoritative(t, r, "www.google.com")
	fromZone := liveFindAuthoritative(t, r, "google.com")

	assert.Equal(t, fromZone, fromHost)
}

func TestFindAuthoritativeNameservers_ReturnsSorted(
	t *testing.T,
) {
	t.Parallel()

	r := newTestResolver(t)
	nameservers := liveFindAuthoritative(t, r, "google.com")

	assert.True(
		t,
		sort.StringsAreSorted(nameservers),
		"nameservers should be sorted, got: %v", nameservers,
	)
}

func TestFindAuthoritativeNameservers_Deterministic(
	t *testing.T,
) {
	t.Parallel()

	r := newTestResolver(t)
	first := liveFindAuthoritative(t, r, "google.com")
	second := liveFindAuthoritative(t, r, "google.com")

	assert.Equal(t, first, second)
}

func TestFindAuthoritativeNameservers_TrailingDot(
	t *testing.T,
) {
	t.Parallel()

	r := newTestResolver(t)
	ns1 := liveFindAuthoritative(t, r, "google.com")
	ns2 := liveFindAuthoritative(t, r, "google.com.")

	assert.Equal(t, ns1, ns2)
}

func TestFindAuthoritativeNameservers_CloudflareDomain(
	t *testing.T,
) {
	t.Parallel()

	r := newTestResolver(t)
	nameservers := liveFindAuthoritative(t, r, "cloudflare.com")

	for _, ns := range nameservers {
		assert.True(t, strings.HasSuffix(ns, "."),
			"NS should be FQDN with trailing dot: %s", ns,
		)
	}
}

// TestResolveNSIPs_EveryNameserver looks up the addresses of two of
// google.com's nameservers together, as the walk does when a referral
// names a zone's nameservers without their addresses, and compares them
// with each looked up alone. Together they must give the addresses of
// both, not only of the first that resolves, so that when one gives no
// usable reply the walk goes on to the other.
func TestResolveNSIPs_EveryNameserver(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	names := []string{"ns3.google.com.", "ns4.google.com."}
	want := make([]string, 0, len(names))

	for _, name := range names {
		want = append(want, liveResolveNSIPs(t, r, []string{name}, 1)...)
	}

	got := liveResolveNSIPs(t, r, names, len(want))

	assert.ElementsMatch(t, want, got)
}

// ----------------------------------------------------------------
// QueryNameserver tests
// ----------------------------------------------------------------

func TestQueryNameserver_BasicA(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ns := findOneNSForDomain(t, r, "google.com")
	resp := liveQueryNameserver(t, r, ns, "www.google.com")

	require.NotNil(t, resp)

	assert.Equal(t, resolver.StatusOK, resp.Status)
	assert.Equal(t, ns, resp.Nameserver)

	hasRecords := len(resp.Records["A"]) > 0 ||
		len(resp.Records["CNAME"]) > 0
	assert.True(t, hasRecords,
		"expected A or CNAME records for www.google.com",
	)
}

func TestQueryNameserver_AAAA(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ns := findOneNSForDomain(t, r, "cloudflare.com")
	resp := liveQueryNameserver(t, r, ns, "cloudflare.com")

	aaaaRecords := resp.Records["AAAA"]
	require.NotEmpty(t, aaaaRecords,
		"cloudflare.com should have AAAA records",
	)

	for _, ip := range aaaaRecords {
		parsed := net.ParseIP(ip)
		require.NotNil(t, parsed,
			"should be valid IP: %s", ip,
		)
	}
}

func TestQueryNameserver_MX(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ns := findOneNSForDomain(t, r, "google.com")
	resp := liveQueryNameserver(t, r, ns, "google.com")

	mxRecords := resp.Records["MX"]
	require.NotEmpty(t, mxRecords,
		"google.com should have MX records",
	)
}

func TestQueryNameserver_TXT(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ns := findOneNSForDomain(t, r, "google.com")
	resp := liveQueryNameserver(t, r, ns, "google.com")

	txtRecords := resp.Records["TXT"]
	require.NotEmpty(t, txtRecords,
		"google.com should have TXT records",
	)

	hasSPF := false

	for _, txt := range txtRecords {
		if strings.Contains(txt, "v=spf1") {
			hasSPF = true

			break
		}
	}

	assert.True(t, hasSPF,
		"google.com should have SPF TXT record",
	)
}

func TestQueryNameserver_NXDomain(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ns := findOneNSForDomain(t, r, "google.com")
	resp := liveQueryNameserver(
		t, r, ns, "this-surely-does-not-exist-xyz.google.com",
	)

	assert.Equal(t, resolver.StatusNXDomain, resp.Status)
}

// TestQueryNameserver_Refused asks a google.com nameserver about
// cloudflare.com, a zone it does not serve, which it refuses. Refusing
// is a failure to answer, not an answer with no records.
func TestQueryNameserver_Refused(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ns := findOneNSForDomain(t, r, "google.com")

	var resp *resolver.NameserverResponse

	livednstest.Retry(
		t,
		"QueryNameserver("+ns+", cloudflare.com)",
		func(ctx context.Context) error {
			var err error

			resp, err = r.QueryNameserver(ctx, ns, "cloudflare.com")
			if err != nil {
				return err
			}

			// A timeout or a network error is no reply at all.
			if resp.Status == resolver.StatusTimeout ||
				strings.HasPrefix(resp.Error, "network error") {
				return fmt.Errorf(
					"%w: %s: %s",
					livednstest.ErrNoAnswer, ns, resp.Error,
				)
			}

			return nil
		},
	)

	assert.Equal(t, resolver.StatusError, resp.Status)
	assert.Equal(t, "server returned REFUSED", resp.Error)
}

func TestQueryNameserver_RecordsSorted(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ns := findOneNSForDomain(t, r, "google.com")
	resp := liveQueryNameserver(t, r, ns, "google.com")

	for recordType, values := range resp.Records {
		assert.True(
			t,
			sort.StringsAreSorted(values),
			"%s records should be sorted", recordType,
		)
	}
}

func TestQueryNameserver_ResponseIncludesNameserver(
	t *testing.T,
) {
	t.Parallel()

	r := newTestResolver(t)
	ns := findOneNSForDomain(t, r, "cloudflare.com")
	resp := liveQueryNameserver(t, r, ns, "cloudflare.com")

	assert.Equal(t, ns, resp.Nameserver)
}

func TestQueryNameserver_EmptyRecordsOnNXDomain(
	t *testing.T,
) {
	t.Parallel()

	r := newTestResolver(t)
	ns := findOneNSForDomain(t, r, "google.com")
	resp := liveQueryNameserver(
		t, r, ns, "this-surely-does-not-exist-xyz.google.com",
	)

	totalRecords := 0
	for _, values := range resp.Records {
		totalRecords += len(values)
	}

	assert.Zero(t, totalRecords)
}

func TestQueryNameserver_TrailingDotHandling(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ns := findOneNSForDomain(t, r, "google.com")
	resp1 := liveQueryNameserver(t, r, ns, "google.com")
	resp2 := liveQueryNameserver(t, r, ns, "google.com.")

	assert.Equal(t, resp1.Status, resp2.Status)
}

// ----------------------------------------------------------------
// QueryAllNameservers tests
// ----------------------------------------------------------------

func TestQueryAllNameservers_ReturnsAllNS(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	results := liveQueryAllNameservers(t, r, "google.com")

	assert.GreaterOrEqual(t, len(results), minNameservers)

	for ns, resp := range results {
		assert.Equal(t, ns, resp.Nameserver)
	}
}

func TestQueryAllNameservers_AllReturnOK(t *testing.T) {
	t.Parallel()

	// The last two names are in zones other than their last two
	// labels: google.co.uk, under the two-label suffix co.uk, and
	// compute-1.amazonaws.com, which amazonaws.com delegates to other
	// servers and which has a host name for each of its addresses.
	// Servers above a name's zone only refer onward, which gives
	// nodata, so ok shows the name was asked at its own zone's
	// servers.
	hostnames := []string{
		"google.com",
		"www.google.co.uk",
		"ec2-3-80-0-1.compute-1.amazonaws.com",
	}

	for _, hostname := range hostnames {
		t.Run(hostname, func(t *testing.T) {
			t.Parallel()

			r := newTestResolver(t)
			results := liveQueryAllNameservers(t, r, hostname)

			// A quorum, not unanimity: one authoritative server
			// being slow or rate-limiting us is a property of the
			// live internet, not a resolver defect.
			assert.GreaterOrEqual(
				t,
				countStatus(results, resolver.StatusOK),
				liveQuorum(len(results)),
				"a quorum of nameservers should answer OK: %s",
				describeStatuses(results),
			)

			// Quorum tolerates SILENCE only. Every individual
			// result must be either the expected answer or a
			// non-answer: ok, timeout or error, and nothing else.
			// Stated as a closed allowlist so that a wrong answer
			// no one thought to ban — nxdomain and nodata today,
			// any status added later — fails here rather than
			// sliding through under the quorum.
			assert.Empty(
				t,
				unsanctionedStatuses(
					results,
					resolver.StatusOK,
					resolver.StatusTimeout,
					resolver.StatusError,
				),
				"every nameserver must answer OK or not answer "+
					"at all: %s",
				describeStatuses(results),
			)
		})
	}
}

func TestQueryAllNameservers_NXDomainFromAllNS(
	t *testing.T,
) {
	t.Parallel()

	r := newTestResolver(t)
	results := liveQueryAllNameservers(
		t, r, "this-surely-does-not-exist-xyz.google.com",
	)

	assert.GreaterOrEqual(
		t,
		countStatus(results, resolver.StatusNXDomain),
		liveQuorum(len(results)),
		"a quorum of nameservers should report NXDOMAIN: %s",
		describeStatuses(results),
	)

	// Silence is tolerated; any actual answer other than NXDOMAIN
	// is not. Closed allowlist for the same reason as above: a
	// server answering `ok` or `nodata` for a name that must not
	// exist is a wrong answer, not a slow one.
	assert.Empty(
		t,
		unsanctionedStatuses(
			results,
			resolver.StatusNXDomain,
			resolver.StatusTimeout,
			resolver.StatusError,
		),
		"every nameserver must report NXDOMAIN or not answer "+
			"at all: %s",
		describeStatuses(results),
	)
}

// ----------------------------------------------------------------
// LookupNS tests
// ----------------------------------------------------------------

func TestLookupNS_ValidDomain(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	nameservers := liveLookupNS(t, r, "google.com")

	for _, ns := range nameservers {
		assert.True(t, strings.HasSuffix(ns, "."),
			"NS should have trailing dot: %s", ns,
		)
	}
}

func TestLookupNS_Sorted(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	nameservers := liveLookupNS(t, r, "google.com")

	assert.True(t, sort.StringsAreSorted(nameservers))
}

func TestLookupNS_MatchesFindAuthoritative(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	fromLookup := liveLookupNS(t, r, "google.com")
	fromFind := liveFindAuthoritative(t, r, "google.com")

	assert.Equal(t, fromFind, fromLookup)
}

// ----------------------------------------------------------------
// ResolveIPAddresses tests
// ----------------------------------------------------------------

func TestResolveIPAddresses_ReturnsIPs(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ips := liveResolveIPs(t, r, "google.com")

	for _, ip := range ips {
		parsed := net.ParseIP(ip)
		assert.NotNil(t, parsed,
			"should be valid IP: %s", ip,
		)
	}
}

func TestResolveIPAddresses_Deduplicated(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ips := liveResolveIPs(t, r, "google.com")

	seen := make(map[string]bool)

	for _, ip := range ips {
		assert.False(t, seen[ip], "duplicate IP: %s", ip)
		seen[ip] = true
	}
}

func TestResolveIPAddresses_Sorted(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ips := liveResolveIPs(t, r, "google.com")

	assert.True(t, sort.StringsAreSorted(ips))
}

func TestResolveIPAddresses_NXDomainReturnsEmpty(
	t *testing.T,
) {
	t.Parallel()

	r := newTestResolver(t)
	ips := liveResolveIPsAllowingEmpty(
		t, r, "this-surely-does-not-exist-xyz.google.com",
	)

	assert.Empty(t, ips)
}

func TestResolveIPAddresses_CloudflareDomain(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ips := liveResolveIPs(t, r, "cloudflare.com")

	assert.NotEmpty(t, ips)
}

// ----------------------------------------------------------------
// Context cancellation tests
// ----------------------------------------------------------------

func TestFindAuthoritativeNameservers_ContextCanceled(
	t *testing.T,
) {
	t.Parallel()

	r := newTestResolver(t)
	ctx, cancel := context.WithCancel(context.Background())
	cancel()

	_, err := r.FindAuthoritativeNameservers(ctx, "google.com")
	assert.Error(t, err)
}

func TestQueryNameserver_ContextCanceled(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ctx, cancel := context.WithCancel(context.Background())
	cancel()

	_, err := r.QueryNameserver(
		ctx, "ns1.google.com.", "google.com",
	)
	assert.Error(t, err)
}

func TestQueryAllNameservers_ContextCanceled(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ctx, cancel := context.WithCancel(context.Background())
	cancel()

	_, err := r.QueryAllNameservers(ctx, "google.com")
	assert.Error(t, err)
}

// TestQueryEachNS_CanceledDuringQuery cancels the context while a
// nameserver is being queried, as shutdown does. A lookup cut short
// says nothing about the nameserver, so it must return an error, not a
// failed response for it.
func TestQueryEachNS_CanceledDuringQuery(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ctx, cancel := context.WithCancel(context.Background())
	t.Cleanup(cancel)

	// Finding the nameserver's address alone starts at the root
	// servers and takes several round trips, so a cancel a few
	// milliseconds in lands during the query.
	time.AfterFunc(5*time.Millisecond, cancel)

	results, err := r.QueryEachNS(
		ctx, []string{"ns1.google.com."}, "google.com",
	)
	require.ErrorIs(t, err, resolver.ErrContextCanceled)
	assert.Nil(t, results)
}

// ----------------------------------------------------------------
// Timeout tests
// ----------------------------------------------------------------

func TestQueryNameserverIP_Timeout(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)

	// Nothing answers at 192.0.2.1, a documentation address. The
	// resolver tries each query twice, and the first try gives up
	// after two seconds. A deadline that ends during the first try
	// makes the status vary from run to run between error and
	// timeout, so the deadline must outlast the first try.
	ctx, cancel := context.WithTimeout(
		context.Background(), 3*time.Second,
	)
	t.Cleanup(cancel)

	resp, err := r.QueryNameserverIP(
		ctx, "unreachable.test.", "192.0.2.1",
		"example.com",
	)
	require.NoError(t, err)

	assert.Equal(t, resolver.StatusTimeout, resp.Status)
	assert.NotEmpty(t, resp.Error)
}

// TestCollectIPs_NoNameserverAnswered takes the response of a
// nameserver at 192.0.2.1, where nothing answers, as
// TestQueryNameserverIP_Timeout does. Addresses collected from
// nameservers that all failed to answer are an error, not none.
func TestCollectIPs_NoNameserverAnswered(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)

	// The deadline outlasts the first try, as in
	// TestQueryNameserverIP_Timeout.
	ctx, cancel := context.WithTimeout(
		context.Background(), 3*time.Second,
	)
	t.Cleanup(cancel)

	resp, err := r.QueryNameserverIP(
		ctx, "unreachable.test.", "192.0.2.1",
		"example.com",
	)
	require.NoError(t, err)

	ips, _, err := resolver.CollectIPs(
		map[string]*resolver.NameserverResponse{resp.Nameserver: resp},
	)
	require.ErrorIs(t, err, resolver.ErrNoNameserverAnswered)
	assert.Empty(t, ips)
}

// TestCollectIPs_ReferralIsNoAnswer asks a root server about
// example.com, which the root zone does not hold, so it only refers the
// query to the com servers. That reply is no answer, as is a parent
// zone's when every server of the name's own zone failed.
func TestCollectIPs_ReferralIsNoAnswer(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)

	var resp *resolver.NameserverResponse

	livednstest.Retry(
		t,
		"QueryNameserverIP(a.root-servers.net, example.com)",
		func(ctx context.Context) error {
			var err error

			resp, err = r.QueryNameserverIP(
				ctx, "a.root-servers.net.", "198.41.0.4",
				"example.com",
			)
			if err != nil {
				return err
			}

			// A timeout or a network error is no reply at all.
			if resp.Status == resolver.StatusTimeout ||
				strings.HasPrefix(resp.Error, "network error") {
				return fmt.Errorf(
					"%w: %s", livednstest.ErrNoAnswer, resp.Error,
				)
			}

			return nil
		},
	)

	assert.Equal(t, resolver.StatusError, resp.Status)
	assert.Equal(t, "server returned a referral", resp.Error)

	ips, _, err := resolver.CollectIPs(
		map[string]*resolver.NameserverResponse{resp.Nameserver: resp},
	)
	require.ErrorIs(t, err, resolver.ErrNoNameserverAnswered)
	assert.Empty(t, ips)
}

func TestResolveIPAddresses_ContextCanceled(t *testing.T) {
	t.Parallel()

	r := newTestResolver(t)
	ctx, cancel := context.WithCancel(context.Background())
	cancel()

	_, err := r.ResolveIPAddresses(ctx, "google.com")
	assert.Error(t, err)
}