feat: add option to log the n most active IPs each hour (#496)

Reviewed-on: https://codeberg.org/Codeberg/pages-server/pulls/496
Co-authored-by: crapStone <me@crapstone.dev>
Co-committed-by: crapStone <me@crapstone.dev>

server/handler/handler.go

@@ -1,8 +1,10 @@
 package handler
 
 import (
+	"net"
 	"net/http"
 	"strings"
+	"sync"
 
 	"github.com/rs/zerolog/log"
 
@@ -24,10 +26,28 @@ func Handler(
 	cfg config.ServerConfig,
 	giteaClient *gitea.Client,
 	canonicalDomainCache, redirectsCache cache.ICache,
+	mostActiveIpMap *sync.Map,
 ) http.HandlerFunc {
 	return func(w http.ResponseWriter, req *http.Request) {
 		ctx := context.New(w, req)
 		log := log.With().Str("ReqId", ctx.ReqId).Strs("Handler", []string{req.Host, req.RequestURI}).Logger()
+
+		if cfg.LogMostActiveIps {
+			ip, _, err := net.SplitHostPort(req.RemoteAddr)
+			if err != nil {
+				log.Warn().Err(err).Msg("Failed to get IP address. Using complete remoteAddr string")
+
+				ip = req.RemoteAddr
+			}
+
+			success := false
+			for !success {
+				value, _ := mostActiveIpMap.LoadOrStore(ip, uint(0))
+				count := value.(uint)
+				success = mostActiveIpMap.CompareAndSwap(ip, count, count+1)
+			}
+		}
+
 		log.Debug().Msg("\n----------------------------------------------------------")
 
 		ctx.RespWriter.Header().Set("Server", "pages-server")

server/handler/handler_test.go

@@ -3,6 +3,7 @@ package handler
 import (
 	"net/http"
 	"net/http/httptest"
+	"sync"
 	"testing"
 	"time"
 
@@ -29,7 +30,7 @@ func TestHandlerPerformance(t *testing.T) {
 		AllowedCorsDomains: []string{"raw.codeberg.org", "fonts.codeberg.org", "design.codeberg.org"},
 		PagesBranches:      []string{"pages"},
 	}
-	testHandler := Handler(serverCfg, giteaClient, cache.NewInMemoryCache(), cache.NewInMemoryCache())
+	testHandler := Handler(serverCfg, giteaClient, cache.NewInMemoryCache(), cache.NewInMemoryCache(), &sync.Map{})
 
 	testCase := func(uri string, status int) {
 		t.Run(uri, func(t *testing.T) {

server/startup.go

@@ -7,7 +7,9 @@ import (
 	"net"
 	"net/http"
 	"os"
+	"sort"
 	"strings"
+	"sync"
 	"time"
 
 	"github.com/pires/go-proxyproto"
@@ -135,8 +137,41 @@ func Serve(ctx *cli.Context) error {
 		StartProfilingServer(ctx.String("profiling-address"))
 	}
 
+	mostActiveIpMap := &sync.Map{}
+	if cfg.Server.LogMostActiveIps {
+		go func() {
+			ticker := time.NewTicker(1 * time.Hour)
+			for range ticker.C {
+				type kv struct {
+					Key   string
+					Value uint
+				}
+
+				var kvArray []kv
+				mostActiveIpMap.Range(func(k, v any) bool {
+					kvArray = append(kvArray, kv{k.(string), v.(uint)})
+					return true
+				})
+				mostActiveIpMap.Clear()
+
+				sort.Slice(kvArray, func(i, j int) bool {
+					return kvArray[i].Value > kvArray[j].Value
+				})
+
+				builder := strings.Builder{}
+				var item kv
+				for i := uint(0); i < cfg.Server.MostActiveIpCount; i++ {
+					item = kvArray[i]
+					builder.WriteString(fmt.Sprintf("\n%s, %d", item.Key, item.Value))
+				}
+
+				log.Log().Msg(fmt.Sprintf("%d most active IPs:%s", cfg.Server.MostActiveIpCount, builder.String()))
+			}
+		}()
+	}
+
 	// Create ssl handler based on settings
-	sslHandler := handler.Handler(cfg.Server, giteaClient, canonicalDomainCache, redirectsCache)
+	sslHandler := handler.Handler(cfg.Server, giteaClient, canonicalDomainCache, redirectsCache, mostActiveIpMap)
 
 	// Start the ssl listener
 	log.Info().Msgf("Start SSL server using TCP listener on %s", listener.Addr())