[优化] 日志查询优化

docs/changelog/index.md

@@ -18,6 +18,9 @@ sidebar: false
 
 ### 变更
 
+- 访问日志列表查询将分页与计数下推到数据库执行，避免百万级数据全量加载到内存。
+- 访问日志 `total_ip` 统计改为 SQL `UNION` + `COUNT(*)` 下推执行，分片计数与分页查询并行化。
+- 访问日志折叠视图、IP 汇总与趋势改为 SQL `GROUP BY` 聚合；过滤条件改为 `node_id` 精确匹配及其他字段前缀匹配以利用索引。
 - 标准化 Server Go 目录结构，引入 `cmd/server`、`openflare-server/internal` 与根级 `pkg` 分层，并拆分原 `utils` 公共能力包。
 
 ## [v2.3.3] - 2026-06-06

openflare-server/.gitignore

@@ -1 +1,2 @@
 /data/
+/postgres-data/

openflare-server/cmd/accesslogbench/main.go

@@ -0,0 +1,562 @@
+package main
+
+import (
+	"flag"
+	"fmt"
+	"log"
+	"os"
+	"path/filepath"
+	"runtime"
+	"sort"
+	"strings"
+	"time"
+
+	"github.com/rain-kl/openflare/openflare-server/internal/common"
+	"github.com/rain-kl/openflare/openflare-server/internal/model"
+)
+
+const retentionDays = 90
+
+func main() {
+	dsn := flag.String("dsn", envOr("DSN", "postgres://openflare:replace-with-strong-password@192.168.107.2:5432/openflare?sslmode=disable"), "PostgreSQL DSN")
+	records := flag.Int("records", 1_000_000, "number of access log rows to seed (0 = skip seeding)")
+	seedNode := flag.String("node-id", "bench-node", "node_id used for seeded rows")
+	iterations := flag.Int("iterations", 10, "benchmark iterations per scenario")
+	warmup := flag.Int("warmup", 2, "warmup iterations per scenario")
+	page := flag.Int("page", 0, "page index for list benchmark")
+	pageSize := flag.Int("page-size", 20, "page size for list benchmark")
+	reset := flag.Bool("reset", false, "truncate node_access_logs shard tables before seeding")
+	legacyIterations := flag.Int("legacy-iterations", 1, "iterations for legacy full-scan scenario (can be very slow)")
+	skipLegacy := flag.Bool("skip-legacy", false, "skip legacy full-scan scenario")
+	verifyOnly := flag.Bool("verify-only", false, "verify query correctness and exit")
+	flag.Parse()
+
+	common.SQLDSN = *dsn
+	common.SQLitePath = filepath.Join(os.TempDir(), "openflare-accesslogbench-missing.sqlite")
+	if err := initBenchDB(*dsn); err != nil {
+		log.Fatalf("init db: %v", err)
+	}
+
+	fmt.Println("=== OpenFlare node_access_logs benchmark (PostgreSQL) ===")
+	fmt.Printf("DSN: %s\n", redactDSN(*dsn))
+	fmt.Printf("GOMAXPROCS=%d\n", runtime.GOMAXPROCS(0))
+
+	if *reset {
+		if err := truncateAccessLogs(); err != nil {
+			log.Fatalf("truncate access logs: %v", err)
+		}
+		fmt.Println("truncated node_access_logs shard tables")
+	}
+
+	if *records > 0 {
+		existing, err := countAllAccessLogs()
+		if err != nil {
+			log.Fatalf("count existing rows: %v", err)
+		}
+		if existing >= int64(*records) {
+			fmt.Printf("existing rows=%d >= target=%d, skip seeding\n", existing, *records)
+		} else {
+			missing := *records - int(existing)
+			fmt.Printf("seeding %d rows (existing=%d)...\n", missing, existing)
+			if err := seedAccessLogs(*seedNode, missing); err != nil {
+				log.Fatalf("seed access logs: %v", err)
+			}
+		}
+	}
+
+	total, err := countAllAccessLogs()
+	if err != nil {
+		log.Fatalf("count rows after seed: %v", err)
+	}
+	fmt.Printf("total rows across shards: %d\n\n", total)
+
+	since := time.Now().UTC().Add(-retentionDays * 24 * time.Hour)
+	if err := verifyQueryCorrectness(since, total); err != nil {
+		log.Fatalf("correctness verification failed: %v", err)
+	}
+	fmt.Println("correctness verification: PASS")
+	if *verifyOnly {
+		return
+	}
+
+	query := model.NodeAccessLogQuery{
+		Since:     since,
+		Page:      *page,
+		PageSize:  *pageSize,
+		SortBy:    "logged_at",
+		SortOrder: "desc",
+	}
+
+	scenarios := []scenario{
+		{
+			name: "paginated_list",
+			run: func() error {
+				_, err := model.ListNodeAccessLogs(query)
+				return err
+			},
+		},
+		{
+			name: "sql_count",
+			run: func() error {
+				_, _, err := model.CountNodeAccessLogs(query)
+				return err
+			},
+		},
+		{
+			name: "api_list+count",
+			run: func() error {
+				if _, err := model.ListNodeAccessLogs(query); err != nil {
+					return err
+				}
+				_, _, err := model.CountNodeAccessLogs(query)
+				return err
+			},
+		},
+		{
+			name: "fullscan_list_legacy",
+			run: func() error {
+				_, err := fullScanList(query)
+				return err
+			},
+		},
+	}
+
+	if err := printExplainPlans(since); err != nil {
+		log.Printf("warn: explain analyze failed: %v", err)
+	}
+
+	for _, item := range scenarios {
+		if item.name == "fullscan_list_legacy" && *skipLegacy {
+			fmt.Println("[fullscan_list_legacy] skipped")
+			continue
+		}
+		iterationCount := *iterations
+		if item.name == "fullscan_list_legacy" {
+			iterationCount = *legacyIterations
+		}
+		result, err := runScenario(item, *warmup, iterationCount)
+		if err != nil {
+			log.Fatalf("scenario %s failed: %v", item.name, err)
+		}
+		printResult(result)
+	}
+}
+
+type scenario struct {
+	name string
+	run  func() error
+}
+
+type benchResult struct {
+	name         string
+	iterations   int
+	latencies    []time.Duration
+	allocBytes   []uint64
+	heapInUse    []uint64
+	maxHeapInUse uint64
+}
+
+func runScenario(item scenario, warmup int, iterations int) (*benchResult, error) {
+	for range warmup {
+		if err := item.run(); err != nil {
+			return nil, err
+		}
+	}
+
+	result := &benchResult{
+		name:       item.name,
+		iterations: iterations,
+	}
+	var peakHeap uint64
+
+	for range iterations {
+		runtime.GC()
+		var before, after runtime.MemStats
+		runtime.ReadMemStats(&before)
+
+		start := time.Now()
+		if err := item.run(); err != nil {
+			return nil, err
+		}
+		elapsed := time.Since(start)
+
+		runtime.ReadMemStats(&after)
+		result.latencies = append(result.latencies, elapsed)
+		result.allocBytes = append(result.allocBytes, after.TotalAlloc-before.TotalAlloc)
+		result.heapInUse = append(result.heapInUse, after.HeapInuse)
+		if after.HeapInuse > peakHeap {
+			peakHeap = after.HeapInuse
+		}
+	}
+	result.maxHeapInUse = peakHeap
+	return result, nil
+}
+
+func printResult(result *benchResult) {
+	latencies := append([]time.Duration(nil), result.latencies...)
+	sort.Slice(latencies, func(i, j int) bool { return latencies[i] < latencies[j] })
+
+	var allocSum uint64
+	var heapSum uint64
+	for index := range result.allocBytes {
+		allocSum += result.allocBytes[index]
+		heapSum += result.heapInUse[index]
+	}
+
+	fmt.Printf("[%s] iterations=%d\n", result.name, result.iterations)
+	fmt.Printf("  latency: min=%s avg=%s p50=%s p95=%s max=%s\n",
+		minDuration(latencies),
+		avgDuration(latencies),
+		percentile(latencies, 50),
+		percentile(latencies, 95),
+		maxDuration(latencies),
+	)
+	fmt.Printf("  alloc/op: avg=%s peak_heap=%s\n",
+		humanBytes(allocSum/uint64(len(result.allocBytes))),
+		humanBytes(result.maxHeapInUse),
+	)
+	fmt.Printf("  heap_inuse/op: avg=%s\n\n", humanBytes(heapSum/uint64(len(result.heapInUse))))
+}
+
+func seedAccessLogs(nodeID string, total int) error {
+	started := time.Now()
+	perShard := total / 10
+	remainder := total % 10
+	seeded := 0
+	now := time.Now().UTC()
+
+	for shard := range 10 {
+		rows := perShard
+		if shard < remainder {
+			rows++
+		}
+		if rows == 0 {
+			continue
+		}
+		table := fmt.Sprintf("node_access_logs_%02d", shard)
+		sql := fmt.Sprintf(`
+INSERT INTO %s (id, node_id, logged_at, remote_addr, region, host, path, status_code, created_at)
+SELECT
+	(gs * 10 + %d)::bigint AS id,
+	$1 AS node_id,
+	$2::timestamptz - (gs || ' minutes')::interval AS logged_at,
+	('203.0.' || ((gs / 256) %% 256)::text || '.' || (gs %% 256)::text) AS remote_addr,
+	'Benchland' AS region,
+	('host-' || (gs %% 200)::text || '.example.com') AS host,
+	('/api/v1/resource/' || gs::text) AS path,
+	(200 + (gs %% 4))::bigint AS status_code,
+	$2::timestamptz AS created_at
+FROM generate_series(0, $3 - 1) AS gs
+`, table, shard)
+		if err := model.DB.Exec(sql, nodeID, now, rows).Error; err != nil {
+			return err
+		}
+		seeded += rows
+		elapsed := time.Since(started)
+		rate := float64(seeded) / elapsed.Seconds()
+		fmt.Printf("  seeded %d/%d rows (%.0f rows/s, elapsed %s)\n", seeded, total, rate, elapsed.Round(time.Millisecond))
+	}
+	return nil
+}
+
+func truncateAccessLogs() error {
+	for _, table := range observabilityShardTables() {
+		if err := model.DB.Exec("TRUNCATE TABLE " + table).Error; err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func countAllAccessLogs() (int64, error) {
+	var total int64
+	for _, table := range observabilityShardTables() {
+		var count int64
+		if err := model.DB.Table(table).Count(&count).Error; err != nil {
+			return 0, err
+		}
+		total += count
+	}
+	return total, nil
+}
+
+func observabilityShardTables() []string {
+	tables := make([]string, 0, 10)
+	for index := range 10 {
+		tables = append(tables, fmt.Sprintf("node_access_logs_%02d", index))
+	}
+	return tables
+}
+
+func verifyQueryCorrectness(since time.Time, tableTotal int64) error {
+	fmt.Println("=== correctness verification ===")
+	baseQuery := model.NodeAccessLogQuery{
+		Since:     since,
+		SortBy:    "logged_at",
+		SortOrder: "desc",
+	}
+
+	reference, err := fullScanList(baseQuery)
+	if err != nil {
+		return fmt.Errorf("reference full scan failed: %w", err)
+	}
+	if int64(len(reference)) != tableTotal {
+		return fmt.Errorf("reference row count %d != table total %d", len(reference), tableTotal)
+	}
+	fmt.Printf("  reference rows in retention window: %d\n", len(reference))
+
+	totalRecords, totalIPs, err := model.CountNodeAccessLogs(baseQuery)
+	if err != nil {
+		return fmt.Errorf("CountNodeAccessLogs failed: %w", err)
+	}
+	if totalRecords != int64(len(reference)) {
+		return fmt.Errorf("total_records=%d want reference=%d", totalRecords, len(reference))
+	}
+	referenceIPs := countUniqueIPs(reference)
+	if totalIPs != referenceIPs {
+		return fmt.Errorf("total_ip=%d want reference=%d", totalIPs, referenceIPs)
+	}
+	fmt.Printf("  count totals: total_record=%d total_ip=%d\n", totalRecords, totalIPs)
+
+	pages := []struct {
+		page     int
+		pageSize int
+	}{
+		{0, 20},
+		{1, 20},
+		{49, 20},
+		{100, 50},
+	}
+	for _, item := range pages {
+		query := baseQuery
+		query.Page = item.page
+		query.PageSize = item.pageSize
+		got, err := model.ListNodeAccessLogs(query)
+		if err != nil {
+			return fmt.Errorf("ListNodeAccessLogs page=%d size=%d failed: %w", item.page, item.pageSize, err)
+		}
+		start := item.page * item.pageSize
+		end := start + item.pageSize
+		if start >= len(reference) {
+			if len(got) != 0 {
+				return fmt.Errorf("page=%d size=%d expected empty got %d", item.page, item.pageSize, len(got))
+			}
+			continue
+		}
+		if end > len(reference) {
+			end = len(reference)
+		}
+		want := reference[start:end]
+		if !accessLogsEqual(got, want) {
+			return fmt.Errorf("page=%d size=%d content mismatch (got %d want %d rows)", item.page, item.pageSize, len(got), len(want))
+		}
+		fmt.Printf("  page=%d size=%d: %d rows match reference\n", item.page, item.pageSize, len(got))
+	}
+
+	filtered := model.NodeAccessLogQuery{
+		NodeID:    "bench-node",
+		Since:     since,
+		SortBy:    "status_code",
+		SortOrder: "asc",
+		Page:      2,
+		PageSize:  15,
+	}
+	filteredReference, err := fullScanList(filtered)
+	if err != nil {
+		return fmt.Errorf("filtered reference failed: %w", err)
+	}
+	filteredRows, err := model.ListNodeAccessLogs(filtered)
+	if err != nil {
+		return fmt.Errorf("filtered ListNodeAccessLogs failed: %w", err)
+	}
+	start := filtered.Page * filtered.PageSize
+	end := start + filtered.PageSize
+	if end > len(filteredReference) {
+		end = len(filteredReference)
+	}
+	if start >= len(filteredReference) {
+		start = len(filteredReference)
+	}
+	if !accessLogsEqual(filteredRows, filteredReference[start:end]) {
+		return fmt.Errorf("filtered page content mismatch")
+	}
+	filteredTotal, filteredIPs, err := model.CountNodeAccessLogs(filtered)
+	if err != nil {
+		return fmt.Errorf("filtered CountNodeAccessLogs failed: %w", err)
+	}
+	if filteredTotal != int64(len(filteredReference)) {
+		return fmt.Errorf("filtered total_records=%d want %d", filteredTotal, len(filteredReference))
+	}
+	if filteredIPs != countUniqueIPs(filteredReference) {
+		return fmt.Errorf("filtered total_ip=%d want %d", filteredIPs, countUniqueIPs(filteredReference))
+	}
+	fmt.Printf("  filtered node_id=bench-node page=2 size=15: match (total_record=%d total_ip=%d)\n", filteredTotal, filteredIPs)
+	return nil
+}
+
+func countUniqueIPs(logs []*model.NodeAccessLog) int64 {
+	ips := make(map[string]struct{})
+	for _, item := range logs {
+		if item == nil {
+			continue
+		}
+		if trimmed := strings.TrimSpace(item.RemoteAddr); trimmed != "" {
+			ips[trimmed] = struct{}{}
+		}
+	}
+	return int64(len(ips))
+}
+
+func accessLogsEqual(left []*model.NodeAccessLog, right []*model.NodeAccessLog) bool {
+	if len(left) != len(right) {
+		return false
+	}
+	for index := range left {
+		if left[index] == nil || right[index] == nil {
+			if left[index] != right[index] {
+				return false
+			}
+			continue
+		}
+		if left[index].ID != right[index].ID ||
+			left[index].NodeID != right[index].NodeID ||
+			!left[index].LoggedAt.Equal(right[index].LoggedAt) ||
+			left[index].RemoteAddr != right[index].RemoteAddr ||
+			left[index].Host != right[index].Host ||
+			left[index].Path != right[index].Path ||
+			left[index].StatusCode != right[index].StatusCode {
+			return false
+		}
+	}
+	return true
+}
+
+func printExplainPlans(since time.Time) error {
+	fmt.Println("=== PostgreSQL EXPLAIN (one shard sample: node_access_logs_00) ===")
+	queries := []struct {
+		name string
+		sql  string
+	}{
+		{
+			name: "paginated_list",
+			sql:  "EXPLAIN (ANALYZE, BUFFERS) SELECT * FROM node_access_logs_00 WHERE logged_at >= $1 ORDER BY logged_at DESC, id DESC LIMIT 20",
+		},
+		{
+			name: "count_rows",
+			sql:  "EXPLAIN (ANALYZE, BUFFERS) SELECT COUNT(*) FROM node_access_logs_00 WHERE logged_at >= $1",
+		},
+		{
+			name: "distinct_ip_union_all",
+			sql: `EXPLAIN (ANALYZE, BUFFERS) SELECT COUNT(*) FROM (
+SELECT remote_addr FROM (
+SELECT TRIM(remote_addr) AS remote_addr FROM node_access_logs_00 WHERE logged_at >= $1 AND remote_addr <> ''
+UNION ALL
+SELECT TRIM(remote_addr) AS remote_addr FROM node_access_logs_01 WHERE logged_at >= $1 AND remote_addr <> ''
+) AS all_ips GROUP BY remote_addr
+) AS ips`,
+		},
+	}
+	for _, item := range queries {
+		rows, err := model.DB.Raw(item.sql, since).Rows()
+		if err != nil {
+			return err
+		}
+		fmt.Printf("-- %s\n", item.name)
+		for rows.Next() {
+			var line string
+			if err := rows.Scan(&line); err != nil {
+				rows.Close()
+				return err
+			}
+			fmt.Println(line)
+		}
+		rows.Close()
+		fmt.Println()
+	}
+	return nil
+}
+
+func fullScanList(query model.NodeAccessLogQuery) ([]*model.NodeAccessLog, error) {
+	legacy := query
+	legacy.PageSize = 0
+	return model.ListNodeAccessLogs(legacy)
+}
+
+func initBenchDB(dsn string) error {
+	return model.InitBenchmarkDB(dsn)
+}
+
+func envOr(key string, fallback string) string {
+	if value := strings.TrimSpace(os.Getenv(key)); value != "" {
+		return value
+	}
+	return fallback
+}
+
+func redactDSN(dsn string) string {
+	if at := strings.Index(dsn, "@"); at > 0 {
+		schemeEnd := strings.Index(dsn, "://")
+		if schemeEnd >= 0 {
+			return dsn[:schemeEnd+3] + "***@" + dsn[at+1:]
+		}
+	}
+	return dsn
+}
+
+func minDuration(values []time.Duration) time.Duration {
+	if len(values) == 0 {
+		return 0
+	}
+	return values[0]
+}
+
+func maxDuration(values []time.Duration) time.Duration {
+	if len(values) == 0 {
+		return 0
+	}
+	return values[len(values)-1]
+}
+
+func avgDuration(values []time.Duration) time.Duration {
+	if len(values) == 0 {
+		return 0
+	}
+	var sum time.Duration
+	for _, value := range values {
+		sum += value
+	}
+	return sum / time.Duration(len(values))
+}
+
+func percentile(values []time.Duration, p int) time.Duration {
+	if len(values) == 0 {
+		return 0
+	}
+	if p <= 0 {
+		return values[0]
+	}
+	if p >= 100 {
+		return values[len(values)-1]
+	}
+	index := (len(values)*p + 99) / 100
+	if index <= 0 {
+		index = 1
+	}
+	if index > len(values) {
+		index = len(values)
+	}
+	return values[index-1]
+}
+
+func humanBytes(value uint64) string {
+	const unit = 1024
+	if value < unit {
+		return fmt.Sprintf("%d B", value)
+	}
+	div, exp := uint64(unit), 0
+	for n := value / unit; n >= unit; n /= unit {
+		div *= unit
+		exp++
+	}
+	return fmt.Sprintf("%.1f %ciB", float64(value)/float64(div), "KMGTPE"[exp])
+}

openflare-server/internal/model/main.go

@@ -336,6 +336,24 @@ func resetPostgresSequence(db *gorm.DB, tableName string) error {
 	return db.Exec(sql).Error
 }
 
+func InitBenchmarkDB(dsn string) error {
+	db, err := gorm.Open(postgres.Open(dsn), &gorm.Config{})
+	if err != nil {
+		return err
+	}
+	sqlDB, err := db.DB()
+	if err != nil {
+		return err
+	}
+	sqlDB.SetMaxOpenConns(20)
+	sqlDB.SetMaxIdleConns(10)
+	DB = db
+	if err = registerSharding(db, "postgres"); err != nil {
+		return err
+	}
+	return ensureDatabaseSchemaUpToDate(db, "postgres")
+}
+
 func InitDB() (err error) {
 	db, backend, err := openDatabase()
 	if err != nil {

openflare-server/internal/model/node_access_log.go

@@ -1,8 +1,10 @@
 package model
 
 import (
+	"fmt"
 	"sort"
 	"strings"
+	"sync"
 	"time"
 
 	"gorm.io/gorm"
@@ -119,15 +121,10 @@ func (log *NodeAccessLog) BeforeCreate(*gorm.DB) error {
 }
 
 func ListNodeAccessLogs(query NodeAccessLogQuery) (logs []*NodeAccessLog, err error) {
-	all, err := listNodeAccessLogsAcrossShards(query)
-	if err != nil {
-		return nil, err
+	if query.PageSize > 0 {
+		return listNodeAccessLogsPaginatedAcrossShards(query)
 	}
-	start, end := paginateBounds(len(all), query.Page, query.PageSize)
-	if start >= len(all) {
-		return []*NodeAccessLog{}, nil
-	}
-	return all[start:end], nil
+	return listNodeAccessLogsAcrossShards(query)
 }
 
 func ListNodeAccessLogsForWAFIPGroup(query NodeAccessLogQuery) ([]*NodeAccessLog, error) {
@@ -135,21 +132,27 @@ func ListNodeAccessLogsForWAFIPGroup(query NodeAccessLogQuery) ([]*NodeAccessLog
 }
 
 func CountNodeAccessLogs(query NodeAccessLogQuery) (totalRecords int64, totalIPs int64, err error) {
-	all, err := listNodeAccessLogsAcrossShards(query)
-	if err != nil {
-		return 0, 0, err
+	db := normalizeShardedDB(DB)
+	var countErr error
+	var distinctErr error
+	var wg sync.WaitGroup
+	wg.Add(2)
+	go func() {
+		defer wg.Done()
+		totalRecords, countErr = countNodeAccessLogRecordsAcrossShards(db, query)
+	}()
+	go func() {
+		defer wg.Done()
+		totalIPs, distinctErr = countDistinctNodeAccessLogIPsAcrossShards(db, query)
+	}()
+	wg.Wait()
+	if countErr != nil {
+		return 0, 0, countErr
 	}
-	ips := make(map[string]struct{}, len(all))
-	for _, item := range all {
-		if item == nil {
-			continue
-		}
-		trimmed := strings.TrimSpace(item.RemoteAddr)
-		if trimmed != "" {
-			ips[trimmed] = struct{}{}
-		}
+	if distinctErr != nil {
+		return 0, 0, distinctErr
 	}
-	return int64(len(all)), int64(len(ips)), nil
+	return totalRecords, totalIPs, nil
 }
 
 func ListNodeAccessLogRegionCounts(nodeID string, since time.Time, limit int) (items []*NodeAccessLogRegionCount, err error) {
@@ -251,38 +254,7 @@ func CountNodeAccessLogIPSummaries(query NodeAccessLogIPSummaryQuery) (total int
 }
 
 func ListNodeAccessLogIPTrend(query NodeAccessLogIPTrendQuery) (items []*NodeAccessLogTrendPointRow, err error) {
-	logs, err := listNodeAccessLogsAcrossShards(NodeAccessLogQuery{
-		NodeID:     query.NodeID,
-		RemoteAddr: query.RemoteAddr,
-		Host:       query.Host,
-		Since:      query.Since,
-	})
-	if err != nil {
-		return nil, err
-	}
-	remoteAddr := strings.TrimSpace(query.RemoteAddr)
-	if remoteAddr == "" {
-		return []*NodeAccessLogTrendPointRow{}, nil
-	}
-	buckets := make(map[int64]int64)
-	for _, item := range logs {
-		if item == nil || strings.TrimSpace(item.RemoteAddr) != remoteAddr {
-			continue
-		}
-		bucketEpoch := bucketEpochForTime(item.LoggedAt, query.BucketMinutes)
-		buckets[bucketEpoch]++
-	}
-	items = make([]*NodeAccessLogTrendPointRow, 0, len(buckets))
-	for bucketEpoch, requestCount := range buckets {
-		items = append(items, &NodeAccessLogTrendPointRow{
-			BucketEpoch:  bucketEpoch,
-			RequestCount: requestCount,
-		})
-	}
-	sort.Slice(items, func(i int, j int) bool {
-		return items[i].BucketEpoch < items[j].BucketEpoch
-	})
-	return items, nil
+	return queryIPTrendRows(query)
 }
 
 func DeleteNodeAccessLogsBefore(before time.Time) (deleted int64, err error) {
@@ -329,26 +301,98 @@ func DeleteNodeAccessLogsByNodeBefore(db *gorm.DB, nodeID string, before time.Ti
 	})
 }
 
-func applyNodeAccessLogFilters(db *gorm.DB, query NodeAccessLogQuery) *gorm.DB {
+func buildNodeAccessLogFilterClause(query NodeAccessLogQuery) (string, []any) {
+	parts := make([]string, 0, 6)
+	args := make([]any, 0, 6)
 	if trimmed := strings.TrimSpace(query.NodeID); trimmed != "" {
-		db = db.Where("node_id LIKE ?", "%"+trimmed+"%")
+		parts = append(parts, "node_id = ?")
+		args = append(args, trimmed)
 	}
 	if trimmed := strings.TrimSpace(query.RemoteAddr); trimmed != "" {
-		db = db.Where("remote_addr LIKE ?", "%"+trimmed+"%")
+		parts = append(parts, "remote_addr LIKE ?")
+		args = append(args, trimmed+"%")
 	}
 	if trimmed := strings.TrimSpace(query.Host); trimmed != "" {
-		db = db.Where("host LIKE ?", "%"+trimmed+"%")
+		parts = append(parts, "host LIKE ?")
+		args = append(args, trimmed+"%")
 	}
 	if trimmed := strings.TrimSpace(query.Path); trimmed != "" {
-		db = db.Where("path LIKE ?", "%"+trimmed+"%")
+		parts = append(parts, "path LIKE ?")
+		args = append(args, trimmed+"%")
 	}
 	if !query.Since.IsZero() {
-		db = db.Where("logged_at >= ?", query.Since)
+		parts = append(parts, "logged_at >= ?")
+		args = append(args, query.Since)
 	}
 	if !query.Until.IsZero() {
-		db = db.Where("logged_at < ?", query.Until)
+		parts = append(parts, "logged_at < ?")
+		args = append(args, query.Until)
 	}
-	return db
+	if len(parts) == 0 {
+		return "TRUE", nil
+	}
+	return strings.Join(parts, " AND "), args
+}
+
+func applyNodeAccessLogFilters(db *gorm.DB, query NodeAccessLogQuery) *gorm.DB {
+	clause, args := buildNodeAccessLogFilterClause(query)
+	if clause == "TRUE" {
+		return db
+	}
+	return db.Where(clause, args...)
+}
+
+func countNodeAccessLogRecordsAcrossShards(db *gorm.DB, query NodeAccessLogQuery) (int64, error) {
+	tables := observabilityShardTables("node_access_logs")
+	counts := make([]int64, len(tables))
+	errs := make([]error, len(tables))
+
+	var wg sync.WaitGroup
+	for index, table := range tables {
+		wg.Add(1)
+		go func(index int, table string) {
+			defer wg.Done()
+			var count int64
+			errs[index] = applyNodeAccessLogFilters(db.Table(table), query).Count(&count).Error
+			counts[index] = count
+		}(index, table)
+	}
+	wg.Wait()
+
+	var total int64
+	for index := range tables {
+		if errs[index] != nil {
+			return 0, errs[index]
+		}
+		total += counts[index]
+	}
+	return total, nil
+}
+
+func countDistinctNodeAccessLogIPsAcrossShards(db *gorm.DB, query NodeAccessLogQuery) (int64, error) {
+	clause, args := buildNodeAccessLogFilterClause(query)
+	tables := observabilityShardTables("node_access_logs")
+	unionParts := make([]string, 0, len(tables))
+	allArgs := make([]any, 0, len(args)*len(tables))
+	for _, table := range tables {
+		unionParts = append(unionParts, fmt.Sprintf(
+			"SELECT TRIM(remote_addr) AS remote_addr FROM %s WHERE %s AND remote_addr <> ''",
+			table,
+			clause,
+		))
+		allArgs = append(allArgs, args...)
+	}
+	sql := fmt.Sprintf(`
+SELECT COUNT(*) FROM (
+	SELECT remote_addr
+	FROM (%s) AS all_ips
+	GROUP BY remote_addr
+) AS ips`, strings.Join(unionParts, " UNION ALL "))
+	var total int64
+	if err := db.Raw(sql, allArgs...).Scan(&total).Error; err != nil {
+		return 0, err
+	}
+	return total, nil
 }
 
 func listNodeAccessLogsAcrossShards(query NodeAccessLogQuery) ([]*NodeAccessLog, error) {
@@ -366,188 +410,59 @@ func listNodeAccessLogsAcrossShards(query NodeAccessLogQuery) ([]*NodeAccessLog,
 	return items, nil
 }
 
-func buildNodeAccessLogBucketRows(query NodeAccessLogBucketQuery) ([]*NodeAccessLogBucketRow, error) {
-	logs, err := listNodeAccessLogsAcrossShards(NodeAccessLogQuery{
-		NodeID:     query.NodeID,
-		RemoteAddr: query.RemoteAddr,
-		Host:       query.Host,
-		Path:       query.Path,
-		Since:      query.Since,
-	})
-	if err != nil {
-		return nil, err
+func listNodeAccessLogsPaginatedAcrossShards(query NodeAccessLogQuery) ([]*NodeAccessLog, error) {
+	fetchLimit := nodeAccessLogFetchLimit(query.Page, query.PageSize)
+	orderClause := nodeAccessLogOrderClause(query.SortBy, query.SortOrder)
+
+	items := make([]*NodeAccessLog, 0, fetchLimit*observabilityShardCount)
+	db := normalizeShardedDB(DB)
+	for _, table := range observabilityShardTables("node_access_logs") {
+		var shardRows []*NodeAccessLog
+		tx := applyNodeAccessLogFilters(db.Table(table), query).Order(orderClause).Limit(fetchLimit)
+		if err := tx.Find(&shardRows).Error; err != nil {
+			return nil, err
+		}
+		items = append(items, shardRows...)
 	}
-	type bucketAccumulator struct {
-		requestCount     int64
-		uniqueIPs        map[string]struct{}
-		uniqueHosts      map[string]struct{}
-		successCount     int64
-		clientErrorCount int64
-		serverErrorCount int64
+
+	sortNodeAccessLogs(items, query.SortBy, query.SortOrder)
+	start, end := paginateBounds(len(items), query.Page, query.PageSize)
+	if start >= len(items) {
+		return []*NodeAccessLog{}, nil
 	}
-	accumulators := make(map[int64]*bucketAccumulator)
-	for _, item := range logs {
-		if item == nil {
-			continue
-		}
-		bucketEpoch := bucketEpochForTime(item.LoggedAt, query.FoldMinutes)
-		accumulator := accumulators[bucketEpoch]
-		if accumulator == nil {
-			accumulator = &bucketAccumulator{
-				uniqueIPs:   make(map[string]struct{}),
-				uniqueHosts: make(map[string]struct{}),
-			}
-			accumulators[bucketEpoch] = accumulator
-		}
-		accumulator.requestCount++
-		if trimmed := strings.TrimSpace(item.RemoteAddr); trimmed != "" {
-			accumulator.uniqueIPs[trimmed] = struct{}{}
-		}
-		if trimmed := strings.TrimSpace(item.Host); trimmed != "" {
-			accumulator.uniqueHosts[trimmed] = struct{}{}
-		}
-		switch {
-		case item.StatusCode < 400:
-			accumulator.successCount++
-		case item.StatusCode < 500:
-			accumulator.clientErrorCount++
-		default:
-			accumulator.serverErrorCount++
-		}
-	}
-	rows := make([]*NodeAccessLogBucketRow, 0, len(accumulators))
-	for bucketEpoch, accumulator := range accumulators {
-		rows = append(rows, &NodeAccessLogBucketRow{
-			BucketEpoch:      bucketEpoch,
-			RequestCount:     accumulator.requestCount,
-			UniqueIPCount:    int64(len(accumulator.uniqueIPs)),
-			UniqueHostCount:  int64(len(accumulator.uniqueHosts)),
-			SuccessCount:     accumulator.successCount,
-			ClientErrorCount: accumulator.clientErrorCount,
-			ServerErrorCount: accumulator.serverErrorCount,
-		})
-	}
-	sortNodeAccessLogBucketRows(rows, query.SortBy, query.SortOrder)
-	return rows, nil
+	return items[start:end], nil
 }
 
-func buildNodeAccessLogBucketIPRows(query NodeAccessLogBucketIPQuery) ([]*NodeAccessLogBucketIPRow, error) {
-	if query.BucketStartedAt.IsZero() {
-		return []*NodeAccessLogBucketIPRow{}, nil
+func nodeAccessLogFetchLimit(page int, pageSize int) int {
+	if page < 0 {
+		page = 0
 	}
-	foldMinutes := query.FoldMinutes
-	if foldMinutes <= 0 {
-		foldMinutes = 3
+	if pageSize <= 0 {
+		return 0
 	}
-	bucketStartedAt := query.BucketStartedAt.UTC()
-	logs, err := listNodeAccessLogsAcrossShards(NodeAccessLogQuery{
-		NodeID:     query.NodeID,
-		RemoteAddr: query.RemoteAddr,
-		Host:       query.Host,
-		Path:       query.Path,
-		Since:      bucketStartedAt,
-		Until:      bucketStartedAt.Add(time.Duration(foldMinutes) * time.Minute),
-	})
-	if err != nil {
-		return nil, err
-	}
-	type accumulator struct {
-		requestCount     int64
-		successCount     int64
-		clientErrorCount int64
-		serverErrorCount int64
-		lastSeenAt       time.Time
-	}
-	accumulators := make(map[string]*accumulator)
-	for _, item := range logs {
-		if item == nil {
-			continue
-		}
-		remoteAddr := strings.TrimSpace(item.RemoteAddr)
-		if remoteAddr == "" {
-			continue
-		}
-		acc := accumulators[remoteAddr]
-		if acc == nil {
-			acc = &accumulator{}
-			accumulators[remoteAddr] = acc
-		}
-		acc.requestCount++
-		switch {
-		case item.StatusCode < 400:
-			acc.successCount++
-		case item.StatusCode < 500:
-			acc.clientErrorCount++
-		default:
-			acc.serverErrorCount++
-		}
-		if item.LoggedAt.After(acc.lastSeenAt) {
-			acc.lastSeenAt = item.LoggedAt
-		}
-	}
-	rows := make([]*NodeAccessLogBucketIPRow, 0, len(accumulators))
-	for remoteAddr, acc := range accumulators {
-		rows = append(rows, &NodeAccessLogBucketIPRow{
-			RemoteAddr:       remoteAddr,
-			RequestCount:     acc.requestCount,
-			SuccessCount:     acc.successCount,
-			ClientErrorCount: acc.clientErrorCount,
-			ServerErrorCount: acc.serverErrorCount,
-			LastSeenEpoch:    acc.lastSeenAt.Unix(),
-		})
-	}
-	sortNodeAccessLogBucketIPRows(rows, query.SortBy, query.SortOrder)
-	return rows, nil
+	return (page + 1) * pageSize
 }
 
-func buildNodeAccessLogIPSummaryRows(query NodeAccessLogIPSummaryQuery, recentSince time.Time) ([]*NodeAccessLogIPSummaryRow, error) {
-	logs, err := listNodeAccessLogsAcrossShards(NodeAccessLogQuery{
-		NodeID:     query.NodeID,
-		RemoteAddr: query.RemoteAddr,
-		Host:       query.Host,
-		Since:      query.Since,
-	})
-	if err != nil {
-		return nil, err
+func nodeAccessLogOrderClause(sortBy string, sortOrder string) string {
+	direction := "DESC"
+	if normalizeSortOrder(sortOrder) == "asc" {
+		direction = "ASC"
 	}
-	type accumulator struct {
-		totalRequests  int64
-		recentRequests int64
-		lastSeenAt     time.Time
+	column := "logged_at"
+	switch strings.TrimSpace(sortBy) {
+	case "status_code":
+		column = "status_code"
+	case "remote_addr":
+		column = "remote_addr"
+	case "host":
+		column = "host"
+	case "path":
+		column = "path"
 	}
-	accumulators := make(map[string]*accumulator)
-	for _, item := range logs {
-		if item == nil {
-			continue
-		}
-		remoteAddr := strings.TrimSpace(item.RemoteAddr)
-		if remoteAddr == "" {
-			continue
-		}
-		acc := accumulators[remoteAddr]
-		if acc == nil {
-			acc = &accumulator{}
-			accumulators[remoteAddr] = acc
-		}
-		acc.totalRequests++
-		if !recentSince.IsZero() && !item.LoggedAt.Before(recentSince) {
-			acc.recentRequests++
-		}
-		if item.LoggedAt.After(acc.lastSeenAt) {
-			acc.lastSeenAt = item.LoggedAt
-		}
+	if column == "logged_at" {
+		return column + " " + direction + ", id " + direction
 	}
-	rows := make([]*NodeAccessLogIPSummaryRow, 0, len(accumulators))
-	for remoteAddr, acc := range accumulators {
-		rows = append(rows, &NodeAccessLogIPSummaryRow{
-			RemoteAddr:     remoteAddr,
-			TotalRequests:  acc.totalRequests,
-			RecentRequests: acc.recentRequests,
-			LastSeenEpoch:  acc.lastSeenAt.Unix(),
-		})
-	}
-	sortNodeAccessLogIPSummaryRows(rows, query.SortBy, query.SortOrder)
-	return rows, nil
+	return column + " " + direction + ", logged_at " + direction + ", id " + direction
 }
 
 func sortNodeAccessLogBucketIPRows(items []*NodeAccessLogBucketIPRow, sortBy string, sortOrder string) {

openflare-server/internal/model/node_access_log_agg.go

@@ -0,0 +1,421 @@
+package model
+
+import (
+	"fmt"
+	"sort"
+	"strings"
+	"time"
+
+	"gorm.io/gorm"
+)
+
+type shardBucketAggregateRow struct {
+	BucketEpoch      int64 `gorm:"column:bucket_epoch"`
+	RequestCount     int64 `gorm:"column:request_count"`
+	SuccessCount     int64 `gorm:"column:success_count"`
+	ClientErrorCount int64 `gorm:"column:client_error_count"`
+	ServerErrorCount int64 `gorm:"column:server_error_count"`
+}
+
+type shardBucketDimensionRow struct {
+	BucketEpoch int64  `gorm:"column:bucket_epoch"`
+	Value       string `gorm:"column:value"`
+}
+
+type shardIPAggregateRow struct {
+	RemoteAddr       string `gorm:"column:remote_addr"`
+	RequestCount     int64  `gorm:"column:request_count"`
+	SuccessCount     int64  `gorm:"column:success_count"`
+	ClientErrorCount int64  `gorm:"column:client_error_count"`
+	ServerErrorCount int64  `gorm:"column:server_error_count"`
+	LastSeenEpoch    int64  `gorm:"column:last_seen_epoch"`
+}
+
+type shardIPSummaryRow struct {
+	RemoteAddr     string `gorm:"column:remote_addr"`
+	TotalRequests  int64  `gorm:"column:total_requests"`
+	RecentRequests int64  `gorm:"column:recent_requests"`
+	LastSeenEpoch  int64  `gorm:"column:last_seen_epoch"`
+}
+
+type shardIPTrendRow struct {
+	BucketEpoch  int64 `gorm:"column:bucket_epoch"`
+	RequestCount int64 `gorm:"column:request_count"`
+}
+
+func buildNodeAccessLogBucketRows(query NodeAccessLogBucketQuery) ([]*NodeAccessLogBucketRow, error) {
+	db := normalizeShardedDB(DB)
+	filter := nodeAccessLogQueryFromBucket(query)
+	clause, args := buildNodeAccessLogFilterClause(filter)
+	bucketSeconds := int64(query.FoldMinutes * 60)
+	if bucketSeconds <= 0 {
+		bucketSeconds = 180
+	}
+	bucketExpr := accessLogBucketEpochExpr(databaseDialect(db), bucketSeconds)
+
+	type bucketAccumulator struct {
+		requestCount     int64
+		uniqueIPs        map[string]struct{}
+		uniqueHosts      map[string]struct{}
+		successCount     int64
+		clientErrorCount int64
+		serverErrorCount int64
+	}
+	accumulators := make(map[int64]*bucketAccumulator)
+
+	for _, table := range observabilityShardTables("node_access_logs") {
+		var partials []shardBucketAggregateRow
+		sql := fmt.Sprintf(`
+SELECT
+	%s AS bucket_epoch,
+	COUNT(*) AS request_count,
+	SUM(CASE WHEN status_code < 400 THEN 1 ELSE 0 END) AS success_count,
+	SUM(CASE WHEN status_code >= 400 AND status_code < 500 THEN 1 ELSE 0 END) AS client_error_count,
+	SUM(CASE WHEN status_code >= 500 THEN 1 ELSE 0 END) AS server_error_count
+FROM %s
+WHERE %s
+GROUP BY bucket_epoch`, bucketExpr, table, clause)
+		if err := db.Raw(sql, args...).Scan(&partials).Error; err != nil {
+			return nil, err
+		}
+		for _, partial := range partials {
+			accumulator := accumulators[partial.BucketEpoch]
+			if accumulator == nil {
+				accumulator = &bucketAccumulator{
+					uniqueIPs:   make(map[string]struct{}),
+					uniqueHosts: make(map[string]struct{}),
+				}
+				accumulators[partial.BucketEpoch] = accumulator
+			}
+			accumulator.requestCount += partial.RequestCount
+			accumulator.successCount += partial.SuccessCount
+			accumulator.clientErrorCount += partial.ClientErrorCount
+			accumulator.serverErrorCount += partial.ServerErrorCount
+		}
+
+		for _, column := range []string{"remote_addr", "host"} {
+			dimensions, err := queryBucketDimensionRows(db, table, clause, args, column, bucketExpr)
+			if err != nil {
+				return nil, err
+			}
+			for _, item := range dimensions {
+				accumulator := accumulators[item.BucketEpoch]
+				if accumulator == nil {
+					accumulator = &bucketAccumulator{
+						uniqueIPs:   make(map[string]struct{}),
+						uniqueHosts: make(map[string]struct{}),
+					}
+					accumulators[item.BucketEpoch] = accumulator
+				}
+				trimmed := strings.TrimSpace(item.Value)
+				if trimmed == "" {
+					continue
+				}
+				switch column {
+				case "remote_addr":
+					accumulator.uniqueIPs[trimmed] = struct{}{}
+				case "host":
+					accumulator.uniqueHosts[trimmed] = struct{}{}
+				}
+			}
+		}
+	}
+
+	rows := make([]*NodeAccessLogBucketRow, 0, len(accumulators))
+	for bucketEpoch, accumulator := range accumulators {
+		rows = append(rows, &NodeAccessLogBucketRow{
+			BucketEpoch:      bucketEpoch,
+			RequestCount:     accumulator.requestCount,
+			UniqueIPCount:    int64(len(accumulator.uniqueIPs)),
+			UniqueHostCount:  int64(len(accumulator.uniqueHosts)),
+			SuccessCount:     accumulator.successCount,
+			ClientErrorCount: accumulator.clientErrorCount,
+			ServerErrorCount: accumulator.serverErrorCount,
+		})
+	}
+	sortNodeAccessLogBucketRows(rows, query.SortBy, query.SortOrder)
+	return rows, nil
+}
+
+func queryBucketDimensionRows(db *gorm.DB, table string, clause string, args []any, column string, bucketExpr string) ([]shardBucketDimensionRow, error) {
+	var rows []shardBucketDimensionRow
+	sql := fmt.Sprintf(`
+SELECT
+	%s AS bucket_epoch,
+	TRIM(%s) AS value
+FROM %s
+WHERE %s AND TRIM(%s) <> ''
+GROUP BY bucket_epoch, TRIM(%s)`, bucketExpr, column, table, clause, column, column)
+	if err := db.Raw(sql, args...).Scan(&rows).Error; err != nil {
+		return nil, err
+	}
+	return rows, nil
+}
+
+func buildNodeAccessLogBucketIPRows(query NodeAccessLogBucketIPQuery) ([]*NodeAccessLogBucketIPRow, error) {
+	if query.BucketStartedAt.IsZero() {
+		return []*NodeAccessLogBucketIPRow{}, nil
+	}
+	foldMinutes := query.FoldMinutes
+	if foldMinutes <= 0 {
+		foldMinutes = 3
+	}
+	bucketStartedAt := query.BucketStartedAt.UTC()
+	filter := NodeAccessLogQuery{
+		NodeID:     query.NodeID,
+		RemoteAddr: query.RemoteAddr,
+		Host:       query.Host,
+		Path:       query.Path,
+		Since:      bucketStartedAt,
+		Until:      bucketStartedAt.Add(time.Duration(foldMinutes) * time.Minute),
+	}
+	rows, err := queryIPAggregateRows(filter, false)
+	if err != nil {
+		return nil, err
+	}
+	sortNodeAccessLogBucketIPRows(rows, query.SortBy, query.SortOrder)
+	return rows, nil
+}
+
+func buildNodeAccessLogIPSummaryRows(query NodeAccessLogIPSummaryQuery, recentSince time.Time) ([]*NodeAccessLogIPSummaryRow, error) {
+	filter := NodeAccessLogQuery{
+		NodeID:     query.NodeID,
+		RemoteAddr: query.RemoteAddr,
+		Host:       query.Host,
+		Since:      query.Since,
+	}
+	db := normalizeShardedDB(DB)
+	clause, args := buildNodeAccessLogFilterClause(filter)
+	lastSeenExpr := accessLogEpochExpr(databaseDialect(db))
+
+	type accumulator struct {
+		totalRequests  int64
+		recentRequests int64
+		lastSeenEpoch  int64
+	}
+	accumulators := make(map[string]*accumulator)
+
+	for _, table := range observabilityShardTables("node_access_logs") {
+		recentClause := "0"
+		queryArgs := make([]any, 0, len(args)+1)
+		if !recentSince.IsZero() {
+			recentClause = "CASE WHEN logged_at >= ? THEN 1 ELSE 0 END"
+			queryArgs = append(queryArgs, recentSince)
+		}
+		queryArgs = append(queryArgs, args...)
+		var partials []shardIPSummaryRow
+		sql := fmt.Sprintf(`
+SELECT
+	TRIM(remote_addr) AS remote_addr,
+	COUNT(*) AS total_requests,
+	SUM(%s) AS recent_requests,
+	MAX(%s) AS last_seen_epoch
+FROM %s
+WHERE %s AND TRIM(remote_addr) <> ''
+GROUP BY TRIM(remote_addr)`, recentClause, lastSeenExpr, table, clause)
+		if err := db.Raw(sql, queryArgs...).Scan(&partials).Error; err != nil {
+			return nil, err
+		}
+		for _, partial := range partials {
+			remoteAddr := strings.TrimSpace(partial.RemoteAddr)
+			if remoteAddr == "" {
+				continue
+			}
+			acc := accumulators[remoteAddr]
+			if acc == nil {
+				acc = &accumulator{}
+				accumulators[remoteAddr] = acc
+			}
+			acc.totalRequests += partial.TotalRequests
+			acc.recentRequests += partial.RecentRequests
+			if partial.LastSeenEpoch > acc.lastSeenEpoch {
+				acc.lastSeenEpoch = partial.LastSeenEpoch
+			}
+		}
+	}
+
+	rows := make([]*NodeAccessLogIPSummaryRow, 0, len(accumulators))
+	for remoteAddr, acc := range accumulators {
+		rows = append(rows, &NodeAccessLogIPSummaryRow{
+			RemoteAddr:     remoteAddr,
+			TotalRequests:  acc.totalRequests,
+			RecentRequests: acc.recentRequests,
+			LastSeenEpoch:  acc.lastSeenEpoch,
+		})
+	}
+	sortNodeAccessLogIPSummaryRows(rows, query.SortBy, query.SortOrder)
+	return rows, nil
+}
+
+func queryIPAggregateRows(filter NodeAccessLogQuery, exactRemoteAddr bool) ([]*NodeAccessLogBucketIPRow, error) {
+	db := normalizeShardedDB(DB)
+	clause, args := buildNodeAccessLogFilterClause(filter)
+	lastSeenExpr := accessLogEpochExpr(databaseDialect(db))
+
+	type accumulator struct {
+		requestCount     int64
+		successCount     int64
+		clientErrorCount int64
+		serverErrorCount int64
+		lastSeenEpoch    int64
+	}
+	accumulators := make(map[string]*accumulator)
+
+	for _, table := range observabilityShardTables("node_access_logs") {
+		queryClause := clause
+		queryArgs := append([]any{}, args...)
+		if exactRemoteAddr {
+			trimmed := strings.TrimSpace(filter.RemoteAddr)
+			if trimmed == "" {
+				return []*NodeAccessLogBucketIPRow{}, nil
+			}
+			queryClause = combineSQLClauses(queryClause, "TRIM(remote_addr) = ?")
+			queryArgs = append(queryArgs, trimmed)
+		}
+		var partials []shardIPAggregateRow
+		sql := fmt.Sprintf(`
+SELECT
+	TRIM(remote_addr) AS remote_addr,
+	COUNT(*) AS request_count,
+	SUM(CASE WHEN status_code < 400 THEN 1 ELSE 0 END) AS success_count,
+	SUM(CASE WHEN status_code >= 400 AND status_code < 500 THEN 1 ELSE 0 END) AS client_error_count,
+	SUM(CASE WHEN status_code >= 500 THEN 1 ELSE 0 END) AS server_error_count,
+	MAX(%s) AS last_seen_epoch
+FROM %s
+WHERE %s AND TRIM(remote_addr) <> ''
+GROUP BY TRIM(remote_addr)`, lastSeenExpr, table, queryClause)
+		if err := db.Raw(sql, queryArgs...).Scan(&partials).Error; err != nil {
+			return nil, err
+		}
+		for _, partial := range partials {
+			remoteAddr := strings.TrimSpace(partial.RemoteAddr)
+			if remoteAddr == "" {
+				continue
+			}
+			acc := accumulators[remoteAddr]
+			if acc == nil {
+				acc = &accumulator{}
+				accumulators[remoteAddr] = acc
+			}
+			acc.requestCount += partial.RequestCount
+			acc.successCount += partial.SuccessCount
+			acc.clientErrorCount += partial.ClientErrorCount
+			acc.serverErrorCount += partial.ServerErrorCount
+			if partial.LastSeenEpoch > acc.lastSeenEpoch {
+				acc.lastSeenEpoch = partial.LastSeenEpoch
+			}
+		}
+	}
+
+	rows := make([]*NodeAccessLogBucketIPRow, 0, len(accumulators))
+	for remoteAddr, acc := range accumulators {
+		rows = append(rows, &NodeAccessLogBucketIPRow{
+			RemoteAddr:       remoteAddr,
+			RequestCount:     acc.requestCount,
+			SuccessCount:     acc.successCount,
+			ClientErrorCount: acc.clientErrorCount,
+			ServerErrorCount: acc.serverErrorCount,
+			LastSeenEpoch:    acc.lastSeenEpoch,
+		})
+	}
+	return rows, nil
+}
+
+func queryIPTrendRows(query NodeAccessLogIPTrendQuery) ([]*NodeAccessLogTrendPointRow, error) {
+	remoteAddr := strings.TrimSpace(query.RemoteAddr)
+	if remoteAddr == "" {
+		return []*NodeAccessLogTrendPointRow{}, nil
+	}
+	db := normalizeShardedDB(DB)
+	filter := NodeAccessLogQuery{
+		NodeID:     query.NodeID,
+		RemoteAddr: remoteAddr,
+		Host:       query.Host,
+		Since:      query.Since,
+	}
+	clause, args := buildNodeAccessLogFilterClause(filter)
+	bucketSeconds := int64(query.BucketMinutes * 60)
+	if bucketSeconds <= 0 {
+		bucketSeconds = 1800
+	}
+	bucketExpr := accessLogBucketEpochExpr(databaseDialect(db), bucketSeconds)
+	queryClause := combineSQLClauses(clause, "TRIM(remote_addr) = ?")
+	queryArgs := append(append([]any{}, args...), remoteAddr)
+
+	buckets := make(map[int64]int64)
+	for _, table := range observabilityShardTables("node_access_logs") {
+		var partials []shardIPTrendRow
+		sql := fmt.Sprintf(`
+SELECT
+	%s AS bucket_epoch,
+	COUNT(*) AS request_count
+FROM %s
+WHERE %s
+GROUP BY bucket_epoch`, bucketExpr, table, queryClause)
+		if err := db.Raw(sql, queryArgs...).Scan(&partials).Error; err != nil {
+			return nil, err
+		}
+		for _, partial := range partials {
+			buckets[partial.BucketEpoch] += partial.RequestCount
+		}
+	}
+
+	items := make([]*NodeAccessLogTrendPointRow, 0, len(buckets))
+	for bucketEpoch, requestCount := range buckets {
+		items = append(items, &NodeAccessLogTrendPointRow{
+			BucketEpoch:  bucketEpoch,
+			RequestCount: requestCount,
+		})
+	}
+	sort.Slice(items, func(i int, j int) bool {
+		return items[i].BucketEpoch < items[j].BucketEpoch
+	})
+	return items, nil
+}
+
+func nodeAccessLogQueryFromBucket(query NodeAccessLogBucketQuery) NodeAccessLogQuery {
+	return NodeAccessLogQuery{
+		NodeID:     query.NodeID,
+		RemoteAddr: query.RemoteAddr,
+		Host:       query.Host,
+		Path:       query.Path,
+		Since:      query.Since,
+	}
+}
+
+func databaseDialect(db *gorm.DB) string {
+	if db == nil || db.Dialector == nil {
+		return "sqlite"
+	}
+	switch db.Dialector.Name() {
+	case "postgres":
+		return "postgres"
+	default:
+		return "sqlite"
+	}
+}
+
+func accessLogBucketEpochExpr(dialect string, bucketSeconds int64) string {
+	switch dialect {
+	case "postgres":
+		return fmt.Sprintf("FLOOR(EXTRACT(EPOCH FROM logged_at AT TIME ZONE 'UTC') / %d) * %d", bucketSeconds, bucketSeconds)
+	default:
+		return fmt.Sprintf("(CAST(strftime('%%s', logged_at) AS INTEGER) / %d) * %d", bucketSeconds, bucketSeconds)
+	}
+}
+
+func accessLogEpochExpr(dialect string) string {
+	switch dialect {
+	case "postgres":
+		return "FLOOR(EXTRACT(EPOCH FROM logged_at AT TIME ZONE 'UTC'))::bigint"
+	default:
+		return "CAST((julianday(logged_at) - 2440587.5) * 86400 AS INTEGER)"
+	}
+}
+
+func combineSQLClauses(left string, right string) string {
+	if strings.TrimSpace(left) == "" || left == "TRUE" {
+		return right
+	}
+	return left + " AND " + right
+}

openflare-server/internal/model/node_access_log_test.go

@@ -0,0 +1,542 @@
+package model
+
+import (
+	"fmt"
+	"sort"
+	"strings"
+	"testing"
+	"time"
+)
+
+func TestListNodeAccessLogsPaginatedAcrossShards(t *testing.T) {
+	db := openBareTestSQLiteDB(t, "node_access_log_pagination.db")
+	if err := registerSharding(db, "sqlite"); err != nil {
+		t.Fatalf("register sharding: %v", err)
+	}
+	if err := autoMigrateAll(db); err != nil {
+		t.Fatalf("auto migrate db: %v", err)
+	}
+	previousDB := DB
+	DB = db
+	t.Cleanup(func() {
+		DB = previousDB
+	})
+
+	now := time.Now().UTC()
+	for index := range 15 {
+		record := &NodeAccessLog{
+			NodeID:     "node-page",
+			LoggedAt:   now.Add(-time.Duration(index) * time.Minute),
+			RemoteAddr: fmt.Sprintf("203.0.113.%d", (index%5)+1),
+			Host:       "example.com",
+			Path:       fmt.Sprintf("/path-%02d", index),
+			StatusCode: 200,
+		}
+		if err := db.Create(record).Error; err != nil {
+			t.Fatalf("seed access log %d: %v", index, err)
+		}
+	}
+
+	query := NodeAccessLogQuery{
+		NodeID:    "node-page",
+		Page:      1,
+		PageSize:  5,
+		SortBy:    "logged_at",
+		SortOrder: "desc",
+	}
+	page, err := ListNodeAccessLogs(query)
+	if err != nil {
+		t.Fatalf("ListNodeAccessLogs failed: %v", err)
+	}
+	if len(page) != 5 {
+		t.Fatalf("expected 5 rows, got %d", len(page))
+	}
+	if page[0].Path != "/path-05" || page[4].Path != "/path-09" {
+		t.Fatalf("unexpected page ordering: %+v", page)
+	}
+
+	totalRecords, totalIPs, err := CountNodeAccessLogs(query)
+	if err != nil {
+		t.Fatalf("CountNodeAccessLogs failed: %v", err)
+	}
+	if totalRecords != 15 {
+		t.Fatalf("expected total_records=15, got %d", totalRecords)
+	}
+	if totalIPs != 5 {
+		t.Fatalf("expected total_ip=5, got %d", totalIPs)
+	}
+}
+
+func TestNodeAccessLogOptimizedQueriesMatchReference(t *testing.T) {
+	db := openBareTestSQLiteDB(t, "node_access_log_correctness.db")
+	if err := registerSharding(db, "sqlite"); err != nil {
+		t.Fatalf("register sharding: %v", err)
+	}
+	if err := autoMigrateAll(db); err != nil {
+		t.Fatalf("auto migrate db: %v", err)
+	}
+	previousDB := DB
+	DB = db
+	t.Cleanup(func() {
+		DB = previousDB
+	})
+
+	now := time.Now().UTC()
+	records := []*NodeAccessLog{
+		{NodeID: "node-a", LoggedAt: now.Add(-5 * time.Minute), RemoteAddr: "1.1.1.1", Host: "a.example.com", Path: "/alpha", StatusCode: 200},
+		{NodeID: "node-a", LoggedAt: now.Add(-4 * time.Minute), RemoteAddr: "2.2.2.2", Host: "a.example.com", Path: "/beta", StatusCode: 404},
+		{NodeID: "node-b", LoggedAt: now.Add(-3 * time.Minute), RemoteAddr: "1.1.1.1", Host: "b.example.com", Path: "/gamma", StatusCode: 502},
+		{NodeID: "node-b", LoggedAt: now.Add(-2 * time.Minute), RemoteAddr: " 3.3.3.3 ", Host: "b.example.com", Path: "/delta", StatusCode: 200},
+		{NodeID: "node-b", LoggedAt: now.Add(-1 * time.Minute), RemoteAddr: "", Host: "b.example.com", Path: "/empty-ip", StatusCode: 200},
+	}
+	for _, record := range records {
+		if err := db.Create(record).Error; err != nil {
+			t.Fatalf("seed access log: %v", err)
+		}
+	}
+
+	baseQuery := NodeAccessLogQuery{
+		Since:     now.Add(-10 * time.Minute),
+		SortBy:    "logged_at",
+		SortOrder: "desc",
+	}
+	reference, err := listNodeAccessLogsAcrossShards(baseQuery)
+	if err != nil {
+		t.Fatalf("reference list failed: %v", err)
+	}
+	referenceTotal, referenceIPs, err := countNodeAccessLogsReference(baseQuery)
+	if err != nil {
+		t.Fatalf("reference count failed: %v", err)
+	}
+
+	totalRecords, totalIPs, err := CountNodeAccessLogs(baseQuery)
+	if err != nil {
+		t.Fatalf("CountNodeAccessLogs failed: %v", err)
+	}
+	if totalRecords != referenceTotal {
+		t.Fatalf("total_records mismatch: got %d want %d", totalRecords, referenceTotal)
+	}
+	if totalIPs != referenceIPs {
+		t.Fatalf("total_ip mismatch: got %d want %d", totalIPs, referenceIPs)
+	}
+	if totalRecords != int64(len(reference)) {
+		t.Fatalf("total_records should equal reference rows: got %d want %d", totalRecords, len(reference))
+	}
+
+	for page := range 3 {
+		query := baseQuery
+		query.Page = page
+		query.PageSize = 2
+		pageRows, err := ListNodeAccessLogs(query)
+		if err != nil {
+			t.Fatalf("ListNodeAccessLogs page %d failed: %v", page, err)
+		}
+		start, end := paginateBounds(len(reference), page, query.PageSize)
+		if start >= len(reference) {
+			if len(pageRows) != 0 {
+				t.Fatalf("page %d expected empty slice, got %d rows", page, len(pageRows))
+			}
+			continue
+		}
+		want := reference[start:end]
+		if !nodeAccessLogsEqual(pageRows, want) {
+			t.Fatalf("page %d mismatch:\n got=%+v\nwant=%+v", page, pageRows, want)
+		}
+	}
+
+	filteredQuery := NodeAccessLogQuery{
+		NodeID:    "node-a",
+		Since:     baseQuery.Since,
+		SortBy:    "status_code",
+		SortOrder: "asc",
+		Page:      0,
+		PageSize:  10,
+	}
+	filteredReference, err := listNodeAccessLogsAcrossShards(filteredQuery)
+	if err != nil {
+		t.Fatalf("filtered reference list failed: %v", err)
+	}
+	filteredRows, err := ListNodeAccessLogs(filteredQuery)
+	if err != nil {
+		t.Fatalf("filtered ListNodeAccessLogs failed: %v", err)
+	}
+	if !nodeAccessLogsEqual(filteredRows, filteredReference) {
+		t.Fatalf("filtered list mismatch:\n got=%+v\nwant=%+v", filteredRows, filteredReference)
+	}
+	filteredTotal, filteredIPs, err := CountNodeAccessLogs(filteredQuery)
+	if err != nil {
+		t.Fatalf("filtered CountNodeAccessLogs failed: %v", err)
+	}
+	wantFilteredTotal, wantFilteredIPs, err := countNodeAccessLogsReference(filteredQuery)
+	if err != nil {
+		t.Fatalf("filtered reference count failed: %v", err)
+	}
+	if filteredTotal != wantFilteredTotal || filteredIPs != wantFilteredIPs {
+		t.Fatalf("filtered count mismatch: got (%d,%d) want (%d,%d)", filteredTotal, filteredIPs, wantFilteredTotal, wantFilteredIPs)
+	}
+}
+
+func countNodeAccessLogsReference(query NodeAccessLogQuery) (int64, int64, error) {
+	all, err := listNodeAccessLogsAcrossShards(query)
+	if err != nil {
+		return 0, 0, err
+	}
+	ips := make(map[string]struct{})
+	for _, item := range all {
+		if item == nil {
+			continue
+		}
+		if trimmed := strings.TrimSpace(item.RemoteAddr); trimmed != "" {
+			ips[trimmed] = struct{}{}
+		}
+	}
+	return int64(len(all)), int64(len(ips)), nil
+}
+
+func nodeAccessLogsEqual(left []*NodeAccessLog, right []*NodeAccessLog) bool {
+	if len(left) != len(right) {
+		return false
+	}
+	for index := range left {
+		if left[index] == nil || right[index] == nil {
+			if left[index] != right[index] {
+				return false
+			}
+			continue
+		}
+		if left[index].ID != right[index].ID ||
+			left[index].NodeID != right[index].NodeID ||
+			!left[index].LoggedAt.Equal(right[index].LoggedAt) ||
+			left[index].RemoteAddr != right[index].RemoteAddr ||
+			left[index].Host != right[index].Host ||
+			left[index].Path != right[index].Path ||
+			left[index].StatusCode != right[index].StatusCode {
+			return false
+		}
+	}
+	return true
+}
+
+func TestNodeAccessLogAggregationsMatchReference(t *testing.T) {
+	db := openBareTestSQLiteDB(t, "node_access_log_agg.db")
+	if err := registerSharding(db, "sqlite"); err != nil {
+		t.Fatalf("register sharding: %v", err)
+	}
+	if err := autoMigrateAll(db); err != nil {
+		t.Fatalf("auto migrate db: %v", err)
+	}
+	previousDB := DB
+	DB = db
+	t.Cleanup(func() {
+		DB = previousDB
+	})
+
+	now := time.Date(2026, 3, 19, 8, 12, 30, 0, time.UTC)
+	records := []*NodeAccessLog{
+		{NodeID: "node-folded", LoggedAt: now.Add(-4 * time.Minute), RemoteAddr: "203.0.113.1", Host: "alpha.example.com", Path: "/first", StatusCode: 200},
+		{NodeID: "node-folded", LoggedAt: now.Add(-3 * time.Minute), RemoteAddr: "203.0.113.1", Host: "alpha.example.com", Path: "/second", StatusCode: 502},
+		{NodeID: "node-folded", LoggedAt: now.Add(-2 * time.Minute), RemoteAddr: "203.0.113.2", Host: "beta.example.com", Path: "/third", StatusCode: 404},
+	}
+	for _, record := range records {
+		if err := db.Create(record).Error; err != nil {
+			t.Fatalf("seed access log: %v", err)
+		}
+	}
+
+	since := now.Add(-10 * time.Minute)
+	bucketRows, err := buildNodeAccessLogBucketRows(NodeAccessLogBucketQuery{
+		NodeID: "node-folded", Since: since, FoldMinutes: 5, SortBy: "request_count", SortOrder: "desc",
+	})
+	if err != nil {
+		t.Fatalf("buildNodeAccessLogBucketRows failed: %v", err)
+	}
+	referenceBuckets := referenceBucketRows(records, 5, "request_count", "desc")
+	if !bucketRowsEqual(bucketRows, referenceBuckets) {
+		t.Fatalf("bucket rows mismatch:\n got=%+v\nwant=%+v", bucketRows, referenceBuckets)
+	}
+
+	if len(bucketRows) == 0 {
+		t.Fatal("expected bucket rows before bucket ip verification")
+	}
+	bucketStartedAt := time.Unix(bucketRows[0].BucketEpoch, 0).UTC()
+	bucketIPRows, err := buildNodeAccessLogBucketIPRows(NodeAccessLogBucketIPQuery{
+		NodeID: "node-folded", BucketStartedAt: bucketStartedAt, FoldMinutes: 5, SortBy: "request_count", SortOrder: "desc",
+	})
+	if err != nil {
+		t.Fatalf("buildNodeAccessLogBucketIPRows failed: %v", err)
+	}
+	referenceBucketIPs := referenceBucketIPRows(records, bucketStartedAt, 5, "request_count", "desc")
+	if !bucketIPRowsEqual(bucketIPRows, referenceBucketIPs) {
+		if len(bucketIPRows) > 0 && len(referenceBucketIPs) > 0 {
+			t.Fatalf("bucket ip rows mismatch:\n got=%+v\nwant=%+v", *bucketIPRows[0], *referenceBucketIPs[0])
+		}
+		t.Fatalf("bucket ip rows mismatch:\n got=%+v\nwant=%+v", bucketIPRows, referenceBucketIPs)
+	}
+
+	recentSince := now.Add(-150 * time.Minute)
+	summaryRows, err := buildNodeAccessLogIPSummaryRows(NodeAccessLogIPSummaryQuery{
+		NodeID: "node-folded", Since: since, SortBy: "total_requests", SortOrder: "desc",
+	}, recentSince)
+	if err != nil {
+		t.Fatalf("buildNodeAccessLogIPSummaryRows failed: %v", err)
+	}
+	referenceSummaries := referenceIPSummaryRows(records, since, recentSince, "total_requests", "desc")
+	if !ipSummaryRowsEqual(summaryRows, referenceSummaries) {
+		t.Fatalf("ip summary rows mismatch:\n got=%+v\nwant=%+v", summaryRows, referenceSummaries)
+	}
+
+	trendRows, err := queryIPTrendRows(NodeAccessLogIPTrendQuery{
+		NodeID: "node-folded", RemoteAddr: "203.0.113.1", Since: since, BucketMinutes: 5,
+	})
+	if err != nil {
+		t.Fatalf("queryIPTrendRows failed: %v", err)
+	}
+	referenceTrend := referenceIPTrendRows(records, "203.0.113.1", 5)
+	if !trendRowsEqual(trendRows, referenceTrend) {
+		t.Fatalf("trend rows mismatch:\n got=%+v\nwant=%+v", trendRows, referenceTrend)
+	}
+}
+
+func referenceBucketRows(records []*NodeAccessLog, foldMinutes int, sortBy string, sortOrder string) []*NodeAccessLogBucketRow {
+	type bucketAccumulator struct {
+		requestCount     int64
+		uniqueIPs        map[string]struct{}
+		uniqueHosts      map[string]struct{}
+		successCount     int64
+		clientErrorCount int64
+		serverErrorCount int64
+	}
+	accumulators := make(map[int64]*bucketAccumulator)
+	for _, item := range records {
+		if item == nil {
+			continue
+		}
+		bucketEpoch := bucketEpochForTime(item.LoggedAt, foldMinutes)
+		accumulator := accumulators[bucketEpoch]
+		if accumulator == nil {
+			accumulator = &bucketAccumulator{
+				uniqueIPs:   make(map[string]struct{}),
+				uniqueHosts: make(map[string]struct{}),
+			}
+			accumulators[bucketEpoch] = accumulator
+		}
+		accumulator.requestCount++
+		if trimmed := strings.TrimSpace(item.RemoteAddr); trimmed != "" {
+			accumulator.uniqueIPs[trimmed] = struct{}{}
+		}
+		if trimmed := strings.TrimSpace(item.Host); trimmed != "" {
+			accumulator.uniqueHosts[trimmed] = struct{}{}
+		}
+		switch {
+		case item.StatusCode < 400:
+			accumulator.successCount++
+		case item.StatusCode < 500:
+			accumulator.clientErrorCount++
+		default:
+			accumulator.serverErrorCount++
+		}
+	}
+	rows := make([]*NodeAccessLogBucketRow, 0, len(accumulators))
+	for bucketEpoch, accumulator := range accumulators {
+		rows = append(rows, &NodeAccessLogBucketRow{
+			BucketEpoch:      bucketEpoch,
+			RequestCount:     accumulator.requestCount,
+			UniqueIPCount:    int64(len(accumulator.uniqueIPs)),
+			UniqueHostCount:  int64(len(accumulator.uniqueHosts)),
+			SuccessCount:     accumulator.successCount,
+			ClientErrorCount: accumulator.clientErrorCount,
+			ServerErrorCount: accumulator.serverErrorCount,
+		})
+	}
+	sortNodeAccessLogBucketRows(rows, sortBy, sortOrder)
+	return rows
+}
+
+func referenceBucketIPRows(records []*NodeAccessLog, bucketStartedAt time.Time, foldMinutes int, sortBy string, sortOrder string) []*NodeAccessLogBucketIPRow {
+	type accumulator struct {
+		requestCount     int64
+		successCount     int64
+		clientErrorCount int64
+		serverErrorCount int64
+		lastSeenAt       time.Time
+	}
+	accumulators := make(map[string]*accumulator)
+	until := bucketStartedAt.Add(time.Duration(foldMinutes) * time.Minute)
+	for _, item := range records {
+		if item == nil || item.LoggedAt.Before(bucketStartedAt) || !item.LoggedAt.Before(until) {
+			continue
+		}
+		remoteAddr := strings.TrimSpace(item.RemoteAddr)
+		if remoteAddr == "" {
+			continue
+		}
+		acc := accumulators[remoteAddr]
+		if acc == nil {
+			acc = &accumulator{}
+			accumulators[remoteAddr] = acc
+		}
+		acc.requestCount++
+		switch {
+		case item.StatusCode < 400:
+			acc.successCount++
+		case item.StatusCode < 500:
+			acc.clientErrorCount++
+		default:
+			acc.serverErrorCount++
+		}
+		if item.LoggedAt.After(acc.lastSeenAt) {
+			acc.lastSeenAt = item.LoggedAt
+		}
+	}
+	rows := make([]*NodeAccessLogBucketIPRow, 0, len(accumulators))
+	for remoteAddr, acc := range accumulators {
+		rows = append(rows, &NodeAccessLogBucketIPRow{
+			RemoteAddr:       remoteAddr,
+			RequestCount:     acc.requestCount,
+			SuccessCount:     acc.successCount,
+			ClientErrorCount: acc.clientErrorCount,
+			ServerErrorCount: acc.serverErrorCount,
+			LastSeenEpoch:    acc.lastSeenAt.Unix(),
+		})
+	}
+	sortNodeAccessLogBucketIPRows(rows, sortBy, sortOrder)
+	return rows
+}
+
+func referenceIPSummaryRows(records []*NodeAccessLog, since time.Time, recentSince time.Time, sortBy string, sortOrder string) []*NodeAccessLogIPSummaryRow {
+	type accumulator struct {
+		totalRequests  int64
+		recentRequests int64
+		lastSeenAt     time.Time
+	}
+	accumulators := make(map[string]*accumulator)
+	for _, item := range records {
+		if item == nil || item.LoggedAt.Before(since) {
+			continue
+		}
+		remoteAddr := strings.TrimSpace(item.RemoteAddr)
+		if remoteAddr == "" {
+			continue
+		}
+		acc := accumulators[remoteAddr]
+		if acc == nil {
+			acc = &accumulator{}
+			accumulators[remoteAddr] = acc
+		}
+		acc.totalRequests++
+		if !recentSince.IsZero() && !item.LoggedAt.Before(recentSince) {
+			acc.recentRequests++
+		}
+		if item.LoggedAt.After(acc.lastSeenAt) {
+			acc.lastSeenAt = item.LoggedAt
+		}
+	}
+	rows := make([]*NodeAccessLogIPSummaryRow, 0, len(accumulators))
+	for remoteAddr, acc := range accumulators {
+		rows = append(rows, &NodeAccessLogIPSummaryRow{
+			RemoteAddr:     remoteAddr,
+			TotalRequests:  acc.totalRequests,
+			RecentRequests: acc.recentRequests,
+			LastSeenEpoch:  acc.lastSeenAt.Unix(),
+		})
+	}
+	sortNodeAccessLogIPSummaryRows(rows, sortBy, sortOrder)
+	return rows
+}
+
+func referenceIPTrendRows(records []*NodeAccessLog, remoteAddr string, bucketMinutes int) []*NodeAccessLogTrendPointRow {
+	buckets := make(map[int64]int64)
+	for _, item := range records {
+		if item == nil || strings.TrimSpace(item.RemoteAddr) != remoteAddr {
+			continue
+		}
+		buckets[bucketEpochForTime(item.LoggedAt, bucketMinutes)]++
+	}
+	rows := make([]*NodeAccessLogTrendPointRow, 0, len(buckets))
+	for bucketEpoch, requestCount := range buckets {
+		rows = append(rows, &NodeAccessLogTrendPointRow{BucketEpoch: bucketEpoch, RequestCount: requestCount})
+	}
+	sort.Slice(rows, func(i int, j int) bool { return rows[i].BucketEpoch < rows[j].BucketEpoch })
+	return rows
+}
+
+func bucketRowsEqual(left []*NodeAccessLogBucketRow, right []*NodeAccessLogBucketRow) bool {
+	if len(left) != len(right) {
+		return false
+	}
+	for index := range left {
+		if left[index] == nil || right[index] == nil {
+			if left[index] != right[index] {
+				return false
+			}
+			continue
+		}
+		if *left[index] != *right[index] {
+			return false
+		}
+	}
+	return true
+}
+
+func bucketIPRowsEqual(left []*NodeAccessLogBucketIPRow, right []*NodeAccessLogBucketIPRow) bool {
+	if len(left) != len(right) {
+		return false
+	}
+	for index := range left {
+		if left[index] == nil || right[index] == nil {
+			if left[index] != right[index] {
+				return false
+			}
+			continue
+		}
+		if *left[index] != *right[index] {
+			return false
+		}
+	}
+	return true
+}
+
+func ipSummaryRowsEqual(left []*NodeAccessLogIPSummaryRow, right []*NodeAccessLogIPSummaryRow) bool {
+	if len(left) != len(right) {
+		return false
+	}
+	for index := range left {
+		if left[index] == nil || right[index] == nil {
+			if left[index] != right[index] {
+				return false
+			}
+			continue
+		}
+		if *left[index] != *right[index] {
+			return false
+		}
+	}
+	return true
+}
+
+func trendRowsEqual(left []*NodeAccessLogTrendPointRow, right []*NodeAccessLogTrendPointRow) bool {
+	if len(left) != len(right) {
+		return false
+	}
+	for index := range left {
+		if left[index] == nil || right[index] == nil {
+			if left[index] != right[index] {
+				return false
+			}
+			continue
+		}
+		if *left[index] != *right[index] {
+			return false
+		}
+	}
+	return true
+}
+
+func TestNodeAccessLogOrderClauseMatchesSort(t *testing.T) {
+	if got := nodeAccessLogOrderClause("logged_at", "desc"); got != "logged_at DESC, id DESC" {
+		t.Fatalf("unexpected logged_at order clause: %q", got)
+	}
+	if got := nodeAccessLogOrderClause("status_code", "asc"); got != "status_code ASC, logged_at ASC, id ASC" {
+		t.Fatalf("unexpected status_code order clause: %q", got)
+	}
+}