perf(clickhouse): P0/P1 access log and WAF query aggregation and SQL pagination

This commit is contained in:
ryan
2026-07-02 15:22:36 +08:00
parent 3e910f1961
commit 6525bef15d
10 changed files with 555 additions and 154 deletions
@@ -5,6 +5,8 @@ package model
import (
"context"
"net"
"net/netip"
"sort"
"strings"
"sync"
@@ -99,12 +101,21 @@ func (s *memoryAccessLogStore) BucketAggregates(_ context.Context, filter OpenFl
s.mu.RLock()
defer s.mu.RUnlock()
rows := s.filterRecords(filter)
aggregates := make(map[int64]*openFlareAccessLogBucketAggregateRow)
type bucketAccumulator struct {
openFlareAccessLogBucketAggregateRow
uniqueIPs map[string]struct{}
uniqueHosts map[string]struct{}
}
aggregates := make(map[int64]*bucketAccumulator)
for _, row := range rows {
bucketEpoch := memoryAccessLogBucketEpoch(row.LoggedAt, bucketSeconds)
item := aggregates[bucketEpoch]
if item == nil {
item = &openFlareAccessLogBucketAggregateRow{BucketEpoch: bucketEpoch}
item = &bucketAccumulator{
openFlareAccessLogBucketAggregateRow: openFlareAccessLogBucketAggregateRow{BucketEpoch: bucketEpoch},
uniqueIPs: make(map[string]struct{}),
uniqueHosts: make(map[string]struct{}),
}
aggregates[bucketEpoch] = item
}
item.RequestCount++
@@ -116,14 +127,61 @@ func (s *memoryAccessLogStore) BucketAggregates(_ context.Context, filter OpenFl
default:
item.ServerErrorCount++
}
if remoteAddr := strings.TrimSpace(row.RemoteAddr); remoteAddr != "" {
item.uniqueIPs[remoteAddr] = struct{}{}
}
if host := strings.TrimSpace(row.Host); host != "" {
item.uniqueHosts[host] = struct{}{}
}
}
result := make([]openFlareAccessLogBucketAggregateRow, 0, len(aggregates))
for _, item := range aggregates {
result = append(result, *item)
item.UniqueIPCount = int64(len(item.uniqueIPs))
item.UniqueHostCount = int64(len(item.uniqueHosts))
result = append(result, item.openFlareAccessLogBucketAggregateRow)
}
bucketRows := make([]*OpenFlareAccessLogBucketRow, len(result))
for index := range result {
bucketRows[index] = &OpenFlareAccessLogBucketRow{
BucketEpoch: result[index].BucketEpoch,
RequestCount: result[index].RequestCount,
UniqueIPCount: result[index].UniqueIPCount,
UniqueHostCount: result[index].UniqueHostCount,
SuccessCount: result[index].SuccessCount,
ClientErrorCount: result[index].ClientErrorCount,
ServerErrorCount: result[index].ServerErrorCount,
}
}
sortOpenFlareAccessLogBucketRows(bucketRows, filter.SortBy, filter.SortOrder)
for index := range result {
result[index] = openFlareAccessLogBucketAggregateRow{
BucketEpoch: bucketRows[index].BucketEpoch,
RequestCount: bucketRows[index].RequestCount,
UniqueIPCount: bucketRows[index].UniqueIPCount,
UniqueHostCount: bucketRows[index].UniqueHostCount,
SuccessCount: bucketRows[index].SuccessCount,
ClientErrorCount: bucketRows[index].ClientErrorCount,
ServerErrorCount: bucketRows[index].ServerErrorCount,
}
}
if filter.PageSize > 0 {
start, end := openFlareAccessLogPaginateBounds(len(result), filter.Page, filter.PageSize)
return result[start:end], nil
}
return result, nil
}
func (s *memoryAccessLogStore) CountBuckets(_ context.Context, filter OpenFlareAccessLogQuery, bucketSeconds int64) (int64, error) {
s.mu.RLock()
defer s.mu.RUnlock()
rows := s.filterRecords(filter)
seen := make(map[int64]struct{})
for _, row := range rows {
seen[memoryAccessLogBucketEpoch(row.LoggedAt, bucketSeconds)] = struct{}{}
}
return int64(len(seen)), nil
}
func (s *memoryAccessLogStore) BucketDimensions(_ context.Context, filter OpenFlareAccessLogQuery, column string, bucketSeconds int64) ([]openFlareAccessLogBucketDimensionRow, error) {
s.mu.RLock()
defer s.mu.RUnlock()
@@ -222,9 +280,91 @@ func (s *memoryAccessLogStore) IPSummaries(_ context.Context, filter OpenFlareAc
item.LastSeenEpoch = epoch
}
}
result := make([]openFlareAccessLogIPSummaryRow, 0, len(aggregates))
summaryRows := make([]*OpenFlareAccessLogIPSummaryRow, 0, len(aggregates))
for _, item := range aggregates {
result = append(result, *item)
summaryRows = append(summaryRows, &OpenFlareAccessLogIPSummaryRow{
RemoteAddr: item.RemoteAddr,
TotalRequests: item.TotalRequests,
RecentRequests: item.RecentRequests,
LastSeenEpoch: item.LastSeenEpoch,
})
}
sortOpenFlareAccessLogIPSummaryRows(summaryRows, filter.SortBy, filter.SortOrder)
if filter.PageSize > 0 {
start, end := openFlareAccessLogPaginateBounds(len(summaryRows), filter.Page, filter.PageSize)
summaryRows = summaryRows[start:end]
}
result := make([]openFlareAccessLogIPSummaryRow, len(summaryRows))
for index, item := range summaryRows {
result[index] = openFlareAccessLogIPSummaryRow{
RemoteAddr: item.RemoteAddr,
TotalRequests: item.TotalRequests,
RecentRequests: item.RecentRequests,
LastSeenEpoch: item.LastSeenEpoch,
}
}
return result, nil
}
func (s *memoryAccessLogStore) CountIPSummaries(_ context.Context, filter OpenFlareAccessLogQuery) (int64, error) {
s.mu.RLock()
defer s.mu.RUnlock()
rows := s.filterRecords(filter)
seen := make(map[string]struct{})
for _, row := range rows {
remoteAddr := strings.TrimSpace(row.RemoteAddr)
if remoteAddr == "" {
continue
}
seen[remoteAddr] = struct{}{}
}
return int64(len(seen)), nil
}
func (s *memoryAccessLogStore) WAFIPAggregates(_ context.Context, filter OpenFlareAccessLogQuery) ([]openFlareAccessLogWAFIPAggregateRow, error) {
s.mu.RLock()
defer s.mu.RUnlock()
rows := s.filterRecords(filter)
aggregates := make(map[string]*openFlareAccessLogWAFIPAggregateRow)
order := make([]string, 0)
for _, row := range rows {
remoteAddr := strings.TrimSpace(row.RemoteAddr)
if remoteAddr == "" {
continue
}
item := aggregates[remoteAddr]
if item == nil {
item = &openFlareAccessLogWAFIPAggregateRow{
RemoteAddr: remoteAddr,
StatusCounts: make(map[int]int64),
}
aggregates[remoteAddr] = item
order = append(order, remoteAddr)
}
item.RequestCount++
item.StatusCounts[row.StatusCode]++
if row.StatusCode == 404 {
item.Status404Count++
}
if row.StatusCode >= 400 && row.StatusCode < 500 {
item.ClientErrorCount++
}
if row.StatusCode >= 500 {
item.ServerErrorCount++
}
if memoryAccessLogHostIsIPLiteral(row.Host) {
item.IPHostCount++
}
epoch := row.LoggedAt.UTC().Unix()
if epoch > item.LastSeenEpoch {
item.LastSeenEpoch = epoch
}
}
result := make([]openFlareAccessLogWAFIPAggregateRow, 0, len(order))
for _, remoteAddr := range order {
if item := aggregates[remoteAddr]; item != nil {
result = append(result, *item)
}
}
return result, nil
}
@@ -324,6 +464,19 @@ func memoryAccessLogMatches(row *OpenFlareAccessLog, query OpenFlareAccessLogQue
return true
}
func memoryAccessLogHostIsIPLiteral(value string) bool {
host := strings.TrimSpace(value)
if host == "" {
return false
}
if parsedHost, _, err := net.SplitHostPort(host); err == nil {
host = parsedHost
}
host = strings.Trim(host, "[]")
_, err := netip.ParseAddr(host)
return err == nil
}
func memoryAccessLogBucketEpoch(loggedAt time.Time, bucketSeconds int64) int64 {
if bucketSeconds <= 0 {
bucketSeconds = 180