Fix err (#2608)

* refactor: refactor workflows contents.

* add tool workflows.

* update field.

* fix: remove chat error.

* Fix err.

* fix error.

* remove cn comment.

* update workflows files.

* update infra config.

* move workflows.

* feat: update bot.

* fix: solve uncorrect outdated msg get.

* update get docIDs logic.

* update

* update skip logic.

* fix

* update.

* fix: delay deleteObject func.

* remove unused content.

* update log type.

* feat: implement request batch count limit.

* update

* update

* feat: add rocksTimeout

* feat: wrap logs

* feat: add logs

* feat: listen config

* feat: enable listen TIME_WAIT port

* feat: add logs

* feat: cache batch

* chore: enable fullUserCache

* feat: push rpc num

* feat: push err

* feat: with operationID

* feat: sleep

* feat: change 1s

* feat: change log

* feat: implement Getbatch in rpcCache.

* feat: print getOnline cost

* feat: change log

* feat: change kafka and push config

* feat: del interface

* feat: fix err

* feat: change config

* feat: go mod

* feat: change config

* feat: change config

* feat: add sleep in push

* feat: warn logs

* feat: logs

* feat: logs

* feat: change port

* feat: start config

* feat: remove port reuse

* feat: prometheus config

* feat: prometheus config

* feat: prometheus config

* feat: add long time send msg to grafana

* feat: init

* feat: init

* feat: implement offline push.

* feat: batch get user online

* feat: implement batch Push spilt

* update go mod

* Revert "feat: change port"

This reverts commit 06d5e944

* feat: change port

* feat: change config

* feat: implement kafka producer and consumer.

* update format,

* add PushMQ log.

* feat: get all online users and init push

* feat: lock in online cache

* feat: config

* fix: init online status

* fix: add logs

* fix: userIDs

* fix: add logs

* feat: update Handler logic.

* update MQ logic.

* update

* update

* fix: method name

* fix: update OfflinePushConsumerHandler.

* fix: prommetrics

* fix: add logs

* fix: ctx

* fix: log

* fix: config

* feat: change port

* fix: atomic online cache status

---------

Co-authored-by: Monet Lee <monet_lee@163.com>

config/openim-api.yml

@@ -11,6 +11,6 @@ prometheus:
   # Whether to enable prometheus
   enable: true
   # Prometheus listening ports, must match the number of api.ports
-  ports: [ 20502 ]
+  ports: [ 12002 ]
   # This address can be accessed via a browser
   grafanaURL: http://127.0.0.1:13000/

config/openim-msggateway.yml

@@ -2,13 +2,13 @@ rpc:
   # The IP address where this RPC service registers itself; if left blank, it defaults to the internal network IP
   registerIP: 
   # List of ports that the RPC service listens on; configuring multiple ports will launch multiple instances. These must match the number of configured prometheus ports
-  ports: [ 10140 ]
+  ports: [ 10140, 10141, 10142, 10143, 10144, 10145, 10146, 10147, 10148, 10149, 10150, 10151, 10152, 10153, 10154, 10155 ]
 
 prometheus:
   # Enable or disable Prometheus monitoring
   enable: true
   # List of ports that Prometheus listens on; these must match the number of rpc.ports to ensure correct monitoring setup
-  ports: [ 20640 ]
+  ports: [ 12140, 12141, 12142, 12143, 12144, 12145, 12146, 12147, 12148, 12149, 12150, 12151, 12152, 12153, 12154, 12155 ]
 
 # IP address that the RPC/WebSocket service listens on; setting to 0.0.0.0 listens on both internal and external IPs. If left blank, it automatically uses the internal network IP
 listenIP: 0.0.0.0
@@ -25,6 +25,3 @@ longConnSvr:
 
 # 1: For Android, iOS, Windows, Mac, and web platforms, only one instance can be online at a time
 multiLoginPolicy: 1
-
-
-

config/openim-msgtransfer.yml

@@ -3,4 +3,4 @@ prometheus:
   enable: true
   # List of ports that Prometheus listens on; each port corresponds to an instance of monitoring. Ensure these are managed accordingly
   # Because four instances have been launched, four ports need to be specified
-  ports: [ 20600, 20601, 20602, 20603 ]
+  ports: [ 12020, 12021, 12022, 12023, 12024, 12025, 12026, 12027 ]

config/openim-push.yml

@@ -1,16 +1,16 @@
 rpc:
   # The IP address where this RPC service registers itself; if left blank, it defaults to the internal network IP
-  registerIP: 
+  registerIP:
   # IP address that the RPC service listens on; setting to 0.0.0.0 listens on both internal and external IPs. If left blank, it automatically uses the internal network IP
   listenIP: 0.0.0.0
   # List of ports that the RPC service listens on; configuring multiple ports will launch multiple instances. These must match the number of configured prometheus ports
-  ports: [ 10170, 10171, 10172, 10173 ]
+  ports: [ 10170, 10171, 10172, 10173, 10174, 10175, 10176, 10177, 10178, 10179, 10180, 10181, 10182, 10183, 10184, 10185 ]
 
 prometheus:
   # Enable or disable Prometheus monitoring
   enable: true
   # List of ports that Prometheus listens on; these must match the number of rpc.ports to ensure correct monitoring setup
-  ports: [ 20670, 20671, 20672, 20673 ]
+  ports: [ 12170, 12171, 12172, 12173, 12174, 12175, 12176, 12177, 12178, 12179, 12180, 12181, 12182, 12183, 12184, 12185 ]
 
 maxConcurrentWorkers: 3
 #Use geTui for offline push notifications, or choose fcm or jpns; corresponding configuration settings must be specified.
@@ -38,9 +38,4 @@ iosPush:
       badgeCount: true
       production: false
 
-
-
-
-
-
-
+fullUserCache: true

config/openim-rpc-auth.yml

@@ -4,15 +4,14 @@ rpc:
   # IP address that the RPC service listens on; setting to 0.0.0.0 listens on both internal and external IPs. If left blank, it automatically uses the internal network IP
   listenIP: 0.0.0.0
   # List of ports that the RPC service listens on; configuring multiple ports will launch multiple instances. These must match the number of configured prometheus ports
-  ports: [ 10160 ]
+  ports: [ 10200 ]
 
 prometheus:
   # Enable or disable Prometheus monitoring
   enable: true
   # List of ports that Prometheus listens on; these must match the number of rpc.ports to ensure correct monitoring setup
-  ports: [ 20660 ]
+  ports: [ 12200 ]
 
 tokenPolicy:
   # Token validity period, in days
   expire: 90
-

config/openim-rpc-conversation.yml

@@ -4,10 +4,10 @@ rpc:
   # IP address that the RPC service listens on; setting to 0.0.0.0 listens on both internal and external IPs. If left blank, it automatically uses the internal network IP
   listenIP: 0.0.0.0
   # List of ports that the RPC service listens on; configuring multiple ports will launch multiple instances. These must match the number of configured prometheus ports
-  ports: [ 10180 ]
+  ports: [ 10220 ]
 
 prometheus:
   # Enable or disable Prometheus monitoring
   enable: true
   # List of ports that Prometheus listens on; these must match the number of rpc.ports to ensure correct monitoring setup
-  ports: [ 20680 ]
+  ports: [ 12220 ]

config/openim-rpc-friend.yml

@@ -4,10 +4,10 @@ rpc:
   # IP address that the RPC service listens on; setting to 0.0.0.0 listens on both internal and external IPs. If left blank, it automatically uses the internal network IP
   listenIP: 0.0.0.0
   # List of ports that the RPC service listens on; configuring multiple ports will launch multiple instances. These must match the number of configured prometheus ports
-  ports: [ 10120 ]
+  ports: [ 10240 ]
 
 prometheus:
   # Enable or disable Prometheus monitoring
   enable: true
   # List of ports that Prometheus listens on; these must match the number of rpc.ports to ensure correct monitoring setup
-  ports: [ 20620 ]
+  ports: [ 12240 ]

config/openim-rpc-group.yml

@@ -4,13 +4,13 @@ rpc:
   # IP address that the RPC service listens on; setting to 0.0.0.0 listens on both internal and external IPs. If left blank, it automatically uses the internal network IP
   listenIP: 0.0.0.0
   # List of ports that the RPC service listens on; configuring multiple ports will launch multiple instances. These must match the number of configured prometheus ports
-  ports: [ 10150 ]
+  ports: [ 10260 ]
 
 prometheus:
   # Enable or disable Prometheus monitoring
   enable: true
   # List of ports that Prometheus listens on; these must match the number of rpc.ports to ensure correct monitoring setup
-  ports: [ 20650 ]
+  ports: [ 12260 ]
 
 
-enableHistoryForNewMembers: true
+enableHistoryForNewMembers: true

config/openim-rpc-msg.yml

@@ -4,17 +4,14 @@ rpc:
   # IP address that the RPC service listens on; setting to 0.0.0.0 listens on both internal and external IPs. If left blank, it automatically uses the internal network IP
   listenIP: 0.0.0.0
   # List of ports that the RPC service listens on; configuring multiple ports will launch multiple instances. These must match the number of configured prometheus ports
-  ports: [ 10130 ]
+  ports: [ 10280 ]
 
 prometheus:
   # Enable or disable Prometheus monitoring
   enable: true
   # List of ports that Prometheus listens on; these must match the number of rpc.ports to ensure correct monitoring setup
-  ports: [ 20630 ]
+  ports: [ 12280 ]
 
 
 # Does sending messages require friend verification
 friendVerify: false
-
-
-

config/openim-rpc-third.yml

@@ -4,13 +4,13 @@ rpc:
   # IP address that the RPC service listens on; setting to 0.0.0.0 listens on both internal and external IPs. If left blank, it automatically uses the internal network IP
   listenIP: 0.0.0.0
   # List of ports that the RPC service listens on; configuring multiple ports will launch multiple instances. These must match the number of configured prometheus ports
-  ports: [ 10190 ]
+  ports: [ 10300 ]
 
 prometheus:
   # Enable or disable Prometheus monitoring
   enable: true
   # List of ports that Prometheus listens on; these must match the number of rpc.ports to ensure correct monitoring setup
-  ports: [ 20690 ]
+  ports: [ 12300 ]
 
 
 object:
@@ -37,4 +37,4 @@ object:
     accessKeyID: 
     accessKeySecret: 
     sessionToken: 
-    publicRead: false
+    publicRead: false

config/openim-rpc-user.yml

@@ -4,14 +4,10 @@ rpc:
   # Listening IP; 0.0.0.0 means both internal and external IPs are listened to, if blank, the internal network IP is automatically obtained by default
   listenIP: 0.0.0.0
   # Listening ports; if multiple are configured, multiple instances will be launched, and must be consistent with the number of prometheus.ports
-  ports: [ 10110 ]
+  ports: [ 10320 ]
 
 prometheus:
   # Whether to enable prometheus
   enable: true
   # Prometheus listening ports, must be consistent with the number of rpc.ports
-  ports: [ 20610 ]
-
-
-
-
+  ports: [ 12320 ]

config/prometheus.yml

@@ -28,56 +28,59 @@ scrape_configs:
       - targets: [ internal_ip:20500 ]
   - job_name: openimserver-openim-api
     static_configs:
-      - targets: [ internal_ip:20502 ]
+      - targets: [ internal_ip:12002 ]
         labels:
           namespace: default
   - job_name: openimserver-openim-msggateway
     static_configs:
-      - targets: [ internal_ip:20640 ]
+      - targets: [ internal_ip:12140 ]
+#      - targets: [ internal_ip:12140, internal_ip:12141, internal_ip:12142, internal_ip:12143, internal_ip:12144, internal_ip:12145, internal_ip:12146, internal_ip:12147, internal_ip:12148, internal_ip:12149, internal_ip:12150, internal_ip:12151, internal_ip:12152, internal_ip:12153, internal_ip:12154, internal_ip:12155 ]
         labels:
           namespace: default
   - job_name: openimserver-openim-msgtransfer
     static_configs:
-      - targets: [ internal_ip:20600, internal_ip:20601, internal_ip:20602, internal_ip:20603 ]
+      - targets: [ internal_ip:12020, internal_ip:12021, internal_ip:12022, internal_ip:12023, internal_ip:12024, internal_ip:12025, internal_ip:12026, internal_ip:12027 ]
+#      - targets: [ internal_ip:12020, internal_ip:12021, internal_ip:12022, internal_ip:12023, internal_ip:12024, internal_ip:12025, internal_ip:12026, internal_ip:12027, internal_ip:12028, internal_ip:12029, internal_ip:12030, internal_ip:12031, internal_ip:12032, internal_ip:12033, internal_ip:12034, internal_ip:12035 ]
         labels:
           namespace: default
   - job_name: openimserver-openim-push
     static_configs:
-      - targets: [ internal_ip:20670, internal_ip:20671, internal_ip:20672, internal_ip:20673]
+      - targets: [ internal_ip:12170, internal_ip:12171, internal_ip:12172, internal_ip:12173, internal_ip:12174, internal_ip:12175, internal_ip:12176, internal_ip:12177 ]
+#      - targets: [ internal_ip:12170, internal_ip:12171, internal_ip:12172, internal_ip:12173, internal_ip:12174, internal_ip:12175, internal_ip:12176, internal_ip:12177, internal_ip:12178, internal_ip:12179, internal_ip:12180, internal_ip:12181, internal_ip:12182, internal_ip:12183, internal_ip:12184, internal_ip:12185 ]
         labels:
           namespace: default
   - job_name: openimserver-openim-rpc-auth
     static_configs:
-      - targets: [ internal_ip:20600 ]
+      - targets: [ internal_ip:12200 ]
         labels:
           namespace: default
   - job_name: openimserver-openim-rpc-conversation
     static_configs:
-      - targets: [ internal_ip:20680 ]
+      - targets: [ internal_ip:12220 ]
         labels:
           namespace: default
   - job_name: openimserver-openim-rpc-friend
     static_configs:
-      - targets: [ internal_ip:20620 ]
+      - targets: [ internal_ip:12240 ]
         labels:
           namespace: default
   - job_name: openimserver-openim-rpc-group
     static_configs:
-      - targets: [ internal_ip:20650 ]
+      - targets: [ internal_ip:12260 ]
         labels:
           namespace: default
   - job_name: openimserver-openim-rpc-msg
     static_configs:
-      - targets: [ internal_ip:20630 ]
+      - targets: [ internal_ip:12280 ]
         labels:
           namespace: default
   - job_name: openimserver-openim-rpc-third
     static_configs:
-      - targets: [ internal_ip:20690 ]
+      - targets: [ internal_ip:12300 ]
         labels:
           namespace: default
   - job_name: openimserver-openim-rpc-user
     static_configs:
-      - targets: [ internal_ip:20610 ]
+      - targets: [ internal_ip:12320 ]
         labels:
           namespace: default

docker-compose.yml

@@ -186,4 +186,3 @@ services:
 #    networks:
 #      - openim
 
-

go.mod

@@ -12,8 +12,8 @@ require (
 	github.com/gorilla/websocket v1.5.1
 	github.com/grpc-ecosystem/go-grpc-prometheus v1.2.0
 	github.com/mitchellh/mapstructure v1.5.0
-	github.com/openimsdk/protocol v0.0.72-alpha.17
-	github.com/openimsdk/tools v0.0.50-alpha.11
+	github.com/openimsdk/protocol v0.0.72-alpha.18
+	github.com/openimsdk/tools v0.0.50-alpha.12
 	github.com/pkg/errors v0.9.1 // indirect
 	github.com/prometheus/client_golang v1.18.0
 	github.com/stretchr/testify v1.9.0

go.sum

@@ -336,10 +336,10 @@ github.com/onsi/gomega v1.25.0 h1:Vw7br2PCDYijJHSfBOWhov+8cAnUf8MfMaIOV323l6Y=
 github.com/onsi/gomega v1.25.0/go.mod h1:r+zV744Re+DiYCIPRlYOTxn0YkOLcAnW8k1xXdMPGhM=
 github.com/openimsdk/gomake v0.0.14-alpha.5 h1:VY9c5x515lTfmdhhPjMvR3BBRrRquAUCFsz7t7vbv7Y=
 github.com/openimsdk/gomake v0.0.14-alpha.5/go.mod h1:PndCozNc2IsQIciyn9mvEblYWZwJmAI+06z94EY+csI=
-github.com/openimsdk/protocol v0.0.72-alpha.17 h1:kB7eyjJHdkc8lpSlLIHskHzbodxkIG4eaK908iQLVdI=
-github.com/openimsdk/protocol v0.0.72-alpha.17/go.mod h1:OZQA9FR55lseYoN2Ql1XAHYKHJGu7OMNkUbuekrKCM8=
-github.com/openimsdk/tools v0.0.50-alpha.11 h1:ClhkRjUVJWbmOiQ14G6do/ES1a6ZueDITv40Apwq/Tc=
-github.com/openimsdk/tools v0.0.50-alpha.11/go.mod h1:h1cYmfyaVtgFbKmb1Cfsl8XwUOMTt8ubVUQrdGtsUh4=
+github.com/openimsdk/protocol v0.0.72-alpha.18 h1:EytTtgZuXMG1cgTlJryqXXSO1J3t3wrLIn3Os2PRBEE=
+github.com/openimsdk/protocol v0.0.72-alpha.18/go.mod h1:OZQA9FR55lseYoN2Ql1XAHYKHJGu7OMNkUbuekrKCM8=
+github.com/openimsdk/tools v0.0.50-alpha.12 h1:rV3BxgqN+F79vZvdoQ+97Eob8ScsRVEM8D+Wrcl23uo=
+github.com/openimsdk/tools v0.0.50-alpha.12/go.mod h1:h1cYmfyaVtgFbKmb1Cfsl8XwUOMTt8ubVUQrdGtsUh4=
 github.com/pelletier/go-toml/v2 v2.1.0 h1:FnwAJ4oYMvbT/34k9zzHuZNrhlz48GB3/s6at6/MHO4=
 github.com/pelletier/go-toml/v2 v2.1.0/go.mod h1:tJU2Z3ZkXwnxa4DPO899bsyIoywizdUvyaeZurnPPDc=
 github.com/pelletier/go-toml/v2 v2.2.2 h1:aYUidT7k73Pcl9nb2gScu7NSrKCSHIDE89b3+6Wq+LM=

internal/msggateway/init.go

@@ -58,7 +58,7 @@ func Start(ctx context.Context, index int, conf *Config) error {
 	)
 
 	hubServer := NewServer(rpcPort, longServer, conf, func(srv *Server) error {
-		longServer.online = rpccache.NewOnlineCache(srv.userRcp, nil, rdb, longServer.subscriberUserOnlineStatusChanges)
+		longServer.online, _ = rpccache.NewOnlineCache(srv.userRcp, nil, rdb, false, longServer.subscriberUserOnlineStatusChanges)
 		return nil
 	})
 

internal/msggateway/ws_server.go

@@ -265,7 +265,7 @@ func (ws *WsServer) registerClient(client *Client) {
 		if clientOK {
 			ws.clients.Set(client.UserID, client)
 			// There is already a connection to the platform
-			log.ZInfo(client.ctx, "repeat login", "userID", client.UserID, "platformID",
+			log.ZDebug(client.ctx, "repeat login", "userID", client.UserID, "platformID",
 				client.PlatformID, "old remote addr", getRemoteAdders(oldClients))
 			ws.onlineUserConnNum.Add(1)
 		} else {
@@ -293,7 +293,7 @@ func (ws *WsServer) registerClient(client *Client) {
 
 	wg.Wait()
 
-	log.ZInfo(
+	log.ZDebug(
 		client.ctx,
 		"user online",
 		"online user Num",
@@ -360,7 +360,7 @@ func (ws *WsServer) unregisterClient(client *Client) {
 	ws.onlineUserConnNum.Add(-1)
 	ws.subscription.DelClient(client)
 	//ws.SetUserOnlineStatus(client.ctx, client, constant.Offline)
-	log.ZInfo(client.ctx, "user offline", "close reason", client.closedErr, "online user Num",
+	log.ZDebug(client.ctx, "user offline", "close reason", client.closedErr, "online user Num",
 		ws.onlineUserNum.Load(), "online user conn Num",
 		ws.onlineUserConnNum.Load(),
 	)

internal/msgtransfer/init.go

@@ -111,7 +111,7 @@ func Start(ctx context.Context, index int, config *Config) error {
 	if err != nil {
 		return err
 	}
-	
+
 	msgTransfer := &MsgTransfer{
 		historyCH:      historyCH,
 		historyMongoCH: historyMongoCH,

internal/msgtransfer/online_history_msg_handler.go

@@ -238,6 +238,7 @@ func (och *OnlineHistoryRedisConsumerHandler) categorizeMessageLists(totalMsgs [
 }
 
 func (och *OnlineHistoryRedisConsumerHandler) handleMsg(ctx context.Context, key, conversationID string, storageList, notStorageList []*ContextMsg) {
+	log.ZInfo(ctx, "handle storage msg")
 	for _, storageMsg := range storageList {
 		log.ZDebug(ctx, "handle storage msg", "msg", storageMsg.message.String())
 	}
@@ -254,16 +255,20 @@ func (och *OnlineHistoryRedisConsumerHandler) handleMsg(ctx context.Context, key
 			log.ZError(ctx, "batch data insert to redis err", err, "storageMsgList", storageMessageList)
 			return
 		}
+		log.ZInfo(ctx, "BatchInsertChat2Cache end")
+
 		if isNewConversation {
 			switch msg.SessionType {
 			case constant.ReadGroupChatType:
-				log.ZInfo(ctx, "group chat first create conversation", "conversationID",
+				log.ZDebug(ctx, "group chat first create conversation", "conversationID",
 					conversationID)
 				userIDs, err := och.groupRpcClient.GetGroupMemberIDs(ctx, msg.GroupID)
 				if err != nil {
 					log.ZWarn(ctx, "get group member ids error", err, "conversationID",
 						conversationID)
 				} else {
+					log.ZInfo(ctx, "GetGroupMemberIDs end")
+
 					if err := och.conversationRpcClient.GroupChatFirstCreateConversation(ctx,
 						msg.GroupID, userIDs); err != nil {
 						log.ZWarn(ctx, "single chat first create conversation error", err,
@@ -282,13 +287,16 @@ func (och *OnlineHistoryRedisConsumerHandler) handleMsg(ctx context.Context, key
 			}
 		}
 
-		log.ZDebug(ctx, "success incr to next topic")
+		log.ZInfo(ctx, "success incr to next topic")
 		err = och.msgTransferDatabase.MsgToMongoMQ(ctx, key, conversationID, storageMessageList, lastSeq)
 		if err != nil {
 			log.ZError(ctx, "Msg To MongoDB MQ error", err, "conversationID",
 				conversationID, "storageList", storageMessageList, "lastSeq", lastSeq)
 		}
+		log.ZInfo(ctx, "MsgToMongoMQ end")
+
 		och.toPushTopic(ctx, key, conversationID, storageList)
+		log.ZInfo(ctx, "toPushTopic end")
 	}
 }
 
@@ -319,7 +327,7 @@ func (och *OnlineHistoryRedisConsumerHandler) handleNotification(ctx context.Con
 func (och *OnlineHistoryRedisConsumerHandler) toPushTopic(ctx context.Context, key, conversationID string, msgs []*ContextMsg) {
 	for _, v := range msgs {
 		log.ZDebug(ctx, "push msg to topic", "msg", v.message.String())
-		och.msgTransferDatabase.MsgToPushMQ(v.ctx, key, conversationID, v.message)
+		_, _, _ = och.msgTransferDatabase.MsgToPushMQ(v.ctx, key, conversationID, v.message)
 	}
 }
 
@@ -344,7 +352,7 @@ func (och *OnlineHistoryRedisConsumerHandler) Cleanup(_ sarama.ConsumerGroupSess
 
 func (och *OnlineHistoryRedisConsumerHandler) ConsumeClaim(session sarama.ConsumerGroupSession,
 	claim sarama.ConsumerGroupClaim) error { // a instance in the consumer group
-	log.ZInfo(context.Background(), "online new session msg come", "highWaterMarkOffset",
+	log.ZDebug(context.Background(), "online new session msg come", "highWaterMarkOffset",
 		claim.HighWaterMarkOffset(), "topic", claim.Topic(), "partition", claim.Partition())
 	och.redisMessageBatches.OnComplete = func(lastMessage *sarama.ConsumerMessage, totalCount int) {
 		session.MarkMessage(lastMessage, "")

internal/msgtransfer/online_msg_to_mongo_handler.go

@@ -57,7 +57,7 @@ func (mc *OnlineHistoryMongoConsumerHandler) handleChatWs2Mongo(ctx context.Cont
 		log.ZError(ctx, "msgFromMQ.MsgData is empty", nil, "cMsg", cMsg)
 		return
 	}
-	log.ZInfo(ctx, "mongo consumer recv msg", "msgs", msgFromMQ.String())
+	log.ZDebug(ctx, "mongo consumer recv msg", "msgs", msgFromMQ.String())
 	err = mc.msgTransferDatabase.BatchInsertChat2DB(ctx, msgFromMQ.ConversationID, msgFromMQ.MsgData, msgFromMQ.LastSeq)
 	if err != nil {
 		log.ZError(

internal/push/offlinepush/dummy/push.go

@@ -28,6 +28,6 @@ type Dummy struct {
 }
 
 func (d *Dummy) Push(ctx context.Context, userIDs []string, title, content string, opts *options.Opts) error {
-	log.ZInfo(ctx, "dummy push")
+	log.ZDebug(ctx, "dummy push")
 	return nil
 }

internal/push/offlinepush/getui/push.go

@@ -23,7 +23,6 @@ import (
 	"time"
 
 	"github.com/openimsdk/open-im-server/v3/internal/push/offlinepush/options"
-
 	"github.com/openimsdk/open-im-server/v3/pkg/common/config"
 	"github.com/openimsdk/open-im-server/v3/pkg/common/storage/cache"
 	"github.com/openimsdk/tools/errs"
@@ -100,7 +99,6 @@ func (g *Client) Push(ctx context.Context, userIDs []string, title, content stri
 						if err = g.batchPush(ctx, token, userIDs[i:end], pushReq); err != nil {
 							log.ZError(ctx, "batchPush failed", err, "index", index, "token", token, "req", pushReq)
 						}
-
 					}
 					if err = g.batchPush(ctx, token, userIDs, pushReq); err != nil {
 						log.ZError(ctx, "batchPush failed", err, "index", index, "token", token, "req", pushReq)

internal/push/offlinepush_handler.go

@@ -63,7 +63,7 @@ func (o *OfflinePushConsumerHandler) handleMsg2OfflinePush(ctx context.Context,
 	}
 }
 
-func (c *OfflinePushConsumerHandler) getOfflinePushInfos(msg *sdkws.MsgData) (title, content string, opts *options.Opts, err error) {
+func (o *OfflinePushConsumerHandler) getOfflinePushInfos(msg *sdkws.MsgData) (title, content string, opts *options.Opts, err error) {
 	type AtTextElem struct {
 		Text       string   `json:"text,omitempty"`
 		AtUserList []string `json:"atUserList,omitempty"`
@@ -108,12 +108,12 @@ func (c *OfflinePushConsumerHandler) getOfflinePushInfos(msg *sdkws.MsgData) (ti
 	return
 }
 
-func (c *OfflinePushConsumerHandler) offlinePushMsg(ctx context.Context, msg *sdkws.MsgData, offlinePushUserIDs []string) error {
-	title, content, opts, err := c.getOfflinePushInfos(msg)
+func (o *OfflinePushConsumerHandler) offlinePushMsg(ctx context.Context, msg *sdkws.MsgData, offlinePushUserIDs []string) error {
+	title, content, opts, err := o.getOfflinePushInfos(msg)
 	if err != nil {
 		return err
 	}
-	err = c.offlinePusher.Push(ctx, offlinePushUserIDs, title, content, opts)
+	err = o.offlinePusher.Push(ctx, offlinePushUserIDs, title, content, opts)
 	if err != nil {
 		prommetrics.MsgOfflinePushFailedCounter.Inc()
 		return err

internal/push/onlinepusher.go

@@ -27,12 +27,12 @@ func newEmptyOnlinePusher() *emptyOnlinePusher {
 
 func (emptyOnlinePusher) GetConnsAndOnlinePush(ctx context.Context, msg *sdkws.MsgData,
 	pushToUserIDs []string) (wsResults []*msggateway.SingleMsgToUserResults, err error) {
-	log.ZWarn(ctx, "emptyOnlinePusher GetConnsAndOnlinePush", nil)
+	log.ZInfo(ctx, "emptyOnlinePusher GetConnsAndOnlinePush", nil)
 	return nil, nil
 }
 func (u emptyOnlinePusher) GetOnlinePushFailedUserIDs(ctx context.Context, msg *sdkws.MsgData,
 	wsResults []*msggateway.SingleMsgToUserResults, pushToUserIDs *[]string) []string {
-	log.ZWarn(ctx, "emptyOnlinePusher GetOnlinePushFailedUserIDs", nil)
+	log.ZInfo(ctx, "emptyOnlinePusher GetOnlinePushFailedUserIDs", nil)
 	return nil
 }
 

internal/push/push_handler.go

@@ -27,6 +27,9 @@ import (
 	"github.com/openimsdk/tools/utils/timeutil"
 	"github.com/redis/go-redis/v9"
 	"google.golang.org/protobuf/proto"
+	"math/rand"
+	"strconv"
+	"time"
 )
 
 type ConsumerHandler struct {
@@ -55,6 +58,7 @@ func NewConsumerHandler(config *Config, database controller.PushDatabase, offlin
 	}
 
 	userRpcClient := rpcclient.NewUserRpcClient(client, config.Share.RpcRegisterName.User, config.Share.IMAdminUserID)
+
 	consumerHandler.offlinePusher = offlinePusher
 	consumerHandler.onlinePusher = NewOnlinePusher(client, config)
 	consumerHandler.groupRpcClient = rpcclient.NewGroupRpcClient(client, config.Share.RpcRegisterName.Group)
@@ -65,7 +69,10 @@ func NewConsumerHandler(config *Config, database controller.PushDatabase, offlin
 	consumerHandler.webhookClient = webhook.NewWebhookClient(config.WebhooksConfig.URL)
 	consumerHandler.config = config
 	consumerHandler.pushDatabase = database
-	consumerHandler.onlineCache = rpccache.NewOnlineCache(userRpcClient, consumerHandler.groupLocalCache, rdb, nil)
+	consumerHandler.onlineCache, err = rpccache.NewOnlineCache(userRpcClient, consumerHandler.groupLocalCache, rdb, config.RpcConfig.FullUserCache, nil)
+	if err != nil {
+		return nil, err
+	}
 	return &consumerHandler, nil
 }
 
@@ -108,6 +115,14 @@ func (*ConsumerHandler) Setup(sarama.ConsumerGroupSession) error { return nil }
 func (*ConsumerHandler) Cleanup(sarama.ConsumerGroupSession) error { return nil }
 
 func (c *ConsumerHandler) ConsumeClaim(sess sarama.ConsumerGroupSession, claim sarama.ConsumerGroupClaim) error {
+	c.onlineCache.Lock.Lock()
+	for c.onlineCache.CurrentPhase.Load() < rpccache.DoSubscribeOver {
+		c.onlineCache.Cond.Wait()
+	}
+	c.onlineCache.Lock.Unlock()
+	ctx := mcontext.SetOperationID(context.TODO(), strconv.FormatInt(time.Now().UnixNano()+int64(rand.Uint32()), 10))
+	log.ZInfo(ctx, "begin consume messages")
+
 	for msg := range claim.Messages() {
 		ctx := c.pushConsumerGroup.GetContextFromMsg(msg)
 		c.handleMs2PsChat(ctx, msg.Value)
@@ -118,20 +133,27 @@ func (c *ConsumerHandler) ConsumeClaim(sess sarama.ConsumerGroupSession, claim s
 
 // Push2User Suitable for two types of conversations, one is SingleChatType and the other is NotificationChatType.
 func (c *ConsumerHandler) Push2User(ctx context.Context, userIDs []string, msg *sdkws.MsgData) (err error) {
-	log.ZDebug(ctx, "Get msg from msg_transfer And push msg", "userIDs", userIDs, "msg", msg.String())
+	log.ZInfo(ctx, "Get msg from msg_transfer And push msg", "userIDs", userIDs, "msg", msg.String())
+	defer func(duration time.Time) {
+		t := time.Since(duration)
+		log.ZInfo(ctx, "Get msg from msg_transfer And push msg", "msg", msg.String(), "time cost", t)
+	}(time.Now())
 	if err := c.webhookBeforeOnlinePush(ctx, &c.config.WebhooksConfig.BeforeOnlinePush, userIDs, msg); err != nil {
 		return err
 	}
+	log.ZInfo(ctx, "webhookBeforeOnlinePush end")
+
 	wsResults, err := c.GetConnsAndOnlinePush(ctx, msg, userIDs)
 	if err != nil {
 		return err
 	}
 
-	log.ZDebug(ctx, "single and notification push result", "result", wsResults, "msg", msg, "push_to_userID", userIDs)
+	log.ZInfo(ctx, "single and notification push result", "result", wsResults, "msg", msg, "push_to_userID", userIDs)
 
 	if !c.shouldPushOffline(ctx, msg) {
 		return nil
 	}
+	log.ZInfo(ctx, "shouldPushOffline end")
 
 	for _, v := range wsResults {
 		//message sender do not need offline push
@@ -150,7 +172,7 @@ func (c *ConsumerHandler) Push2User(ctx context.Context, userIDs []string, msg *
 		offlinePushUserID, msg, nil); err != nil {
 		return err
 	}
-
+	log.ZInfo(ctx, "webhookBeforeOfflinePush end")
 	err = c.offlinePushMsg(ctx, msg, offlinePushUserID)
 	if err != nil {
 		log.ZWarn(ctx, "offlinePushMsg failed", err, "offlinePushUserID", offlinePushUserID, "msg", msg)
@@ -172,21 +194,11 @@ func (c *ConsumerHandler) shouldPushOffline(_ context.Context, msg *sdkws.MsgDat
 }
 
 func (c *ConsumerHandler) GetConnsAndOnlinePush(ctx context.Context, msg *sdkws.MsgData, pushToUserIDs []string) ([]*msggateway.SingleMsgToUserResults, error) {
-	var (
-		onlineUserIDs  []string
-		offlineUserIDs []string
-	)
-	for _, userID := range pushToUserIDs {
-		online, err := c.onlineCache.GetUserOnline(ctx, userID)
-		if err != nil {
-			return nil, err
-		}
-		if online {
-			onlineUserIDs = append(onlineUserIDs, userID)
-		} else {
-			offlineUserIDs = append(offlineUserIDs, userID)
-		}
+	onlineUserIDs, offlineUserIDs, err := c.onlineCache.GetUsersOnline(ctx, pushToUserIDs)
+	if err != nil {
+		return nil, err
 	}
+
 	log.ZDebug(ctx, "GetConnsAndOnlinePush online cache", "sendID", msg.SendID, "recvID", msg.RecvID, "groupID", msg.GroupID, "sessionType", msg.SessionType, "clientMsgID", msg.ClientMsgID, "serverMsgID", msg.ServerMsgID, "offlineUserIDs", offlineUserIDs, "onlineUserIDs", onlineUserIDs)
 	var result []*msggateway.SingleMsgToUserResults
 	if len(onlineUserIDs) > 0 {
@@ -205,35 +217,42 @@ func (c *ConsumerHandler) GetConnsAndOnlinePush(ctx context.Context, msg *sdkws.
 }
 
 func (c *ConsumerHandler) Push2Group(ctx context.Context, groupID string, msg *sdkws.MsgData) (err error) {
-	log.ZDebug(ctx, "Get group msg from msg_transfer and push msg", "msg", msg.String(), "groupID", groupID)
+	log.ZInfo(ctx, "Get group msg from msg_transfer and push msg", "msg", msg.String(), "groupID", groupID)
+	defer func(duration time.Time) {
+		t := time.Since(duration)
+		log.ZInfo(ctx, "Get group msg from msg_transfer and push msg end", "msg", msg.String(), "groupID", groupID, "time cost", t)
+	}(time.Now())
 	var pushToUserIDs []string
 	if err = c.webhookBeforeGroupOnlinePush(ctx, &c.config.WebhooksConfig.BeforeGroupOnlinePush, groupID, msg,
 		&pushToUserIDs); err != nil {
 		return err
 	}
+	log.ZInfo(ctx, "webhookBeforeGroupOnlinePush end")
 
 	err = c.groupMessagesHandler(ctx, groupID, &pushToUserIDs, msg)
 	if err != nil {
 		return err
 	}
+	log.ZInfo(ctx, "groupMessagesHandler end")
 
 	wsResults, err := c.GetConnsAndOnlinePush(ctx, msg, pushToUserIDs)
 	if err != nil {
 		return err
 	}
 
-	log.ZDebug(ctx, "group push result", "result", wsResults, "msg", msg)
+	log.ZInfo(ctx, "group push result", "result", wsResults, "msg", msg)
 
 	if !c.shouldPushOffline(ctx, msg) {
 		return nil
 	}
 	needOfflinePushUserIDs := c.onlinePusher.GetOnlinePushFailedUserIDs(ctx, msg, wsResults, &pushToUserIDs)
-
+	log.ZInfo(ctx, "GetOnlinePushFailedUserIDs end")
 	//filter some user, like don not disturb or don't need offline push etc.
 	needOfflinePushUserIDs, err = c.filterGroupMessageOfflinePush(ctx, groupID, msg, needOfflinePushUserIDs)
 	if err != nil {
 		return err
 	}
+	log.ZInfo(ctx, "filterGroupMessageOfflinePush end")
 
 	// Use offline push messaging
 	if len(needOfflinePushUserIDs) > 0 {
@@ -295,7 +314,7 @@ func (c *ConsumerHandler) groupMessagesHandler(ctx context.Context, groupID stri
 				if unmarshalNotificationElem(msg.Content, &tips) != nil {
 					return err
 				}
-				log.ZInfo(ctx, "GroupDismissedNotificationInfo****", "groupID", groupID, "num", len(*pushToUserIDs), "list", pushToUserIDs)
+				log.ZDebug(ctx, "GroupDismissedNotificationInfo****", "groupID", groupID, "num", len(*pushToUserIDs), "list", pushToUserIDs)
 				if len(c.config.Share.IMAdminUserID) > 0 {
 					ctx = mcontext.WithOpUserIDContext(ctx, c.config.Share.IMAdminUserID[0])
 				}
@@ -379,6 +398,7 @@ func (c *ConsumerHandler) getOfflinePushInfos(msg *sdkws.MsgData) (title, conten
 	}
 	return
 }
+
 func (c *ConsumerHandler) DeleteMemberAndSetConversationSeq(ctx context.Context, groupID string, userIDs []string) error {
 	conversationID := msgprocessor.GetConversationIDBySessionType(constant.ReadGroupChatType, groupID)
 	maxSeq, err := c.msgRpcClient.GetConversationMaxSeq(ctx, conversationID)
@@ -387,6 +407,7 @@ func (c *ConsumerHandler) DeleteMemberAndSetConversationSeq(ctx context.Context,
 	}
 	return c.conversationRpcClient.SetConversationMaxSeq(ctx, userIDs, conversationID, maxSeq)
 }
+
 func unmarshalNotificationElem(bytes []byte, t any) error {
 	var notification sdkws.NotificationElem
 	if err := json.Unmarshal(bytes, &notification); err != nil {

internal/rpc/msg/clear.go

@@ -67,7 +67,7 @@ func (m *msgServer) ClearMsg(ctx context.Context, req *msg.ClearMsgReq) (_ *msg.
 		return nil, err
 	}
 
-	log.ZInfo(ctx, "clearing message", "docNum", docNum, "msgNum", msgNum, "cost", time.Since(start))
+	log.ZDebug(ctx, "clearing message", "docNum", docNum, "msgNum", msgNum, "cost", time.Since(start))
 
 	return &msg.ClearMsgResp{}, nil
 }

internal/rpc/user/online.go

@@ -2,6 +2,8 @@ package user
 
 import (
 	"context"
+	"github.com/openimsdk/tools/utils/datautil"
+
 	"github.com/openimsdk/protocol/constant"
 	pbuser "github.com/openimsdk/protocol/user"
 )
@@ -80,3 +82,22 @@ func (s *userServer) SetUserOnlineStatus(ctx context.Context, req *pbuser.SetUse
 	}
 	return &pbuser.SetUserOnlineStatusResp{}, nil
 }
+
+func (s *userServer) GetAllOnlineUsers(ctx context.Context, req *pbuser.GetAllOnlineUsersReq) (*pbuser.GetAllOnlineUsersResp, error) {
+	resMap, nextCursor, err := s.online.GetAllOnlineUsers(ctx, req.Cursor)
+	if err != nil {
+		return nil, err
+	}
+	resp := &pbuser.GetAllOnlineUsersResp{
+		StatusList: make([]*pbuser.OnlineStatus, 0, len(resMap)),
+		NextCursor: nextCursor,
+	}
+	for userID, plats := range resMap {
+		resp.StatusList = append(resp.StatusList, &pbuser.OnlineStatus{
+			UserID:      userID,
+			Status:      int32(datautil.If(len(plats) > 0, constant.Online, constant.Offline)),
+			PlatformIDs: plats,
+		})
+	}
+	return resp, nil
+}

internal/tools/cron_task.go

@@ -79,13 +79,13 @@ func Start(ctx context.Context, config *CronTaskConfig) error {
 		now := time.Now()
 		deltime := now.Add(-time.Hour * 24 * time.Duration(config.CronTask.RetainChatRecords))
 		ctx := mcontext.SetOperationID(ctx, fmt.Sprintf("cron_%d_%d", os.Getpid(), deltime.UnixMilli()))
-		log.ZInfo(ctx, "clear chat records", "deltime", deltime, "timestamp", deltime.UnixMilli())
+		log.ZDebug(ctx, "clear chat records", "deltime", deltime, "timestamp", deltime.UnixMilli())
 
 		if _, err := msgClient.ClearMsg(ctx, &msg.ClearMsgReq{Timestamp: deltime.UnixMilli()}); err != nil {
 			log.ZError(ctx, "cron clear chat records failed", err, "deltime", deltime, "cont", time.Since(now))
 			return
 		}
-		log.ZInfo(ctx, "cron clear chat records success", "deltime", deltime, "cont", time.Since(now))
+		log.ZDebug(ctx, "cron clear chat records success", "deltime", deltime, "cont", time.Since(now))
 	}
 	if _, err := crontab.AddFunc(config.CronTask.CronExecuteTime, clearMsgFunc); err != nil {
 		return errs.Wrap(err)
@@ -95,7 +95,7 @@ func Start(ctx context.Context, config *CronTaskConfig) error {
 	msgDestructFunc := func() {
 		now := time.Now()
 		ctx := mcontext.SetOperationID(ctx, fmt.Sprintf("cron_%d_%d", os.Getpid(), now.UnixMilli()))
-		log.ZInfo(ctx, "msg destruct cron start", "now", now)
+		log.ZDebug(ctx, "msg destruct cron start", "now", now)
 
 		conversations, err := conversationClient.GetConversationsNeedDestructMsgs(ctx, &pbconversation.GetConversationsNeedDestructMsgsReq{})
 		if err != nil {
@@ -108,7 +108,7 @@ func Start(ctx context.Context, config *CronTaskConfig) error {
 				return
 			}
 		}
-		log.ZInfo(ctx, "msg destruct cron task completed", "cont", time.Since(now))
+		log.ZDebug(ctx, "msg destruct cron task completed", "cont", time.Since(now))
 	}
 	if _, err := crontab.AddFunc(config.CronTask.CronExecuteTime, msgDestructFunc); err != nil {
 		return errs.Wrap(err)
@@ -119,18 +119,18 @@ func Start(ctx context.Context, config *CronTaskConfig) error {
 	// 	now := time.Now()
 	// 	deleteTime := now.Add(-time.Hour * 24 * time.Duration(config.CronTask.FileExpireTime))
 	// 	ctx := mcontext.SetOperationID(ctx, fmt.Sprintf("cron_%d_%d", os.Getpid(), deleteTime.UnixMilli()))
-	// 	log.ZInfo(ctx, "deleteoutDatedData ", "deletetime", deleteTime, "timestamp", deleteTime.UnixMilli())
+	// 	log.ZDebug(ctx, "deleteoutDatedData ", "deletetime", deleteTime, "timestamp", deleteTime.UnixMilli())
 	// 	if _, err := thirdClient.DeleteOutdatedData(ctx, &third.DeleteOutdatedDataReq{ExpireTime: deleteTime.UnixMilli()}); err != nil {
 	// 		log.ZError(ctx, "cron deleteoutDatedData failed", err, "deleteTime", deleteTime, "cont", time.Since(now))
 	// 		return
 	// 	}
-	// 	log.ZInfo(ctx, "cron deleteoutDatedData success", "deltime", deleteTime, "cont", time.Since(now))
+	// 	log.ZDebug(ctx, "cron deleteoutDatedData success", "deltime", deleteTime, "cont", time.Since(now))
 	// }
 	// if _, err := crontab.AddFunc(config.CronTask.CronExecuteTime, deleteObjectFunc); err != nil {
 	// 	return errs.Wrap(err)
 	// }
 
-	log.ZInfo(ctx, "start cron task", "CronExecuteTime", config.CronTask.CronExecuteTime)
+	log.ZDebug(ctx, "start cron task", "CronExecuteTime", config.CronTask.CronExecuteTime)
 	crontab.Start()
 	<-ctx.Done()
 	return nil

pkg/common/config/config.go

@@ -224,6 +224,7 @@ type Push struct {
 		BadgeCount bool   `mapstructure:"badgeCount"`
 		Production bool   `mapstructure:"production"`
 	} `mapstructure:"iosPush"`
+	FullUserCache bool `mapstructure:"fullUserCache"`
 }
 
 type Auth struct {

pkg/common/startrpc/start.go

@@ -54,15 +54,11 @@ func Start[T any](ctx context.Context, discovery *config.Discovery, prometheusCo
 	log.CInfo(ctx, "RPC server is initializing", "rpcRegisterName", rpcRegisterName, "rpcPort", rpcPort,
 		"prometheusPorts", prometheusConfig.Ports)
 	rpcTcpAddr := net.JoinHostPort(network.GetListenIP(listenIP), strconv.Itoa(rpcPort))
+
 	listener, err := net.Listen(
 		"tcp",
 		rpcTcpAddr,
 	)
-	if err != nil {
-		return errs.WrapMsg(err, "listen err", "rpcTcpAddr", rpcTcpAddr)
-	}
-
-	defer listener.Close()
 	client, err := kdisc.NewDiscoveryRegister(discovery, share)
 	if err != nil {
 		return err

pkg/common/storage/cache/cachekey/online.go

@@ -1,6 +1,9 @@
 package cachekey
 
-import "time"
+import (
+	"strings"
+	"time"
+)
 
 const (
 	OnlineKey     = "ONLINE:"
@@ -11,3 +14,7 @@ const (
 func GetOnlineKey(userID string) string {
 	return OnlineKey + userID
 }
+
+func GetOnlineKeyUserID(key string) string {
+	return strings.TrimPrefix(key, OnlineKey)
+}

pkg/common/storage/cache/online.go

@@ -5,4 +5,5 @@ import "context"
 type OnlineCache interface {
 	GetOnline(ctx context.Context, userID string) ([]int32, error)
 	SetUserOnline(ctx context.Context, userID string, online, offline []int32) error
+	GetAllOnlineUsers(ctx context.Context, cursor uint64) (map[string][]int32, uint64, error)
 }

pkg/common/storage/cache/redis/batch.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"encoding/json"
 	"github.com/dtm-labs/rockscache"
+	"github.com/openimsdk/tools/errs"
 	"github.com/openimsdk/tools/log"
 	"github.com/redis/go-redis/v9"
 	"golang.org/x/sync/singleflight"
@@ -65,6 +66,7 @@ func batchGetCache2[K comparable, V any](ctx context.Context, rcClient *rockscac
 				}
 				bs, err := json.Marshal(value)
 				if err != nil {
+					log.ZError(ctx, "marshal failed", err)
 					return nil, err
 				}
 				cacheIndex[index] = string(bs)
@@ -72,7 +74,7 @@ func batchGetCache2[K comparable, V any](ctx context.Context, rcClient *rockscac
 			return cacheIndex, nil
 		})
 		if err != nil {
-			return nil, err
+			return nil, errs.WrapMsg(err, "FetchBatch2 failed")
 		}
 		for index, data := range indexCache {
 			if data == "" {
@@ -80,7 +82,7 @@ func batchGetCache2[K comparable, V any](ctx context.Context, rcClient *rockscac
 			}
 			var value V
 			if err := json.Unmarshal([]byte(data), &value); err != nil {
-				return nil, err
+				return nil, errs.WrapMsg(err, "Unmarshal failed")
 			}
 			if cb, ok := any(&value).(BatchCacheCallback[K]); ok {
 				cb.BatchCache(keyId[keys[index]])

pkg/common/storage/cache/redis/batch_handler.go

@@ -28,6 +28,10 @@ import (
 	"time"
 )
 
+const (
+	rocksCacheTimeout = 11 * time.Second
+)
+
 // BatchDeleterRedis is a concrete implementation of the BatchDeleter interface based on Redis and RocksCache.
 type BatchDeleterRedis struct {
 	redisClient    redis.UniversalClient
@@ -106,6 +110,8 @@ func (c *BatchDeleterRedis) AddKeys(keys ...string) {
 // GetRocksCacheOptions returns the default configuration options for RocksCache.
 func GetRocksCacheOptions() *rockscache.Options {
 	opts := rockscache.NewDefaultOptions()
+	opts.LockExpire = rocksCacheTimeout
+	opts.WaitReplicasTimeout = rocksCacheTimeout
 	opts.StrongConsistency = true
 	opts.RandomExpireAdjustment = 0.2
 

pkg/common/storage/cache/redis/online.go

@@ -2,8 +2,10 @@ package redis
 
 import (
 	"context"
+	"fmt"
 	"github.com/openimsdk/open-im-server/v3/pkg/common/storage/cache"
 	"github.com/openimsdk/open-im-server/v3/pkg/common/storage/cache/cachekey"
+	"github.com/openimsdk/protocol/constant"
 	"github.com/openimsdk/tools/errs"
 	"github.com/openimsdk/tools/log"
 	"github.com/redis/go-redis/v9"
@@ -49,6 +51,36 @@ func (s *userOnline) GetOnline(ctx context.Context, userID string) ([]int32, err
 	return platformIDs, nil
 }
 
+func (s *userOnline) GetAllOnlineUsers(ctx context.Context, cursor uint64) (map[string][]int32, uint64, error) {
+	result := make(map[string][]int32)
+
+	keys, nextCursor, err := s.rdb.Scan(ctx, cursor, fmt.Sprintf("%s*", cachekey.OnlineKey), constant.ParamMaxLength).Result()
+	if err != nil {
+		return nil, 0, err
+	}
+
+	for _, key := range keys {
+		userID := cachekey.GetOnlineKeyUserID(key)
+		strValues, err := s.rdb.ZRange(ctx, key, 0, -1).Result()
+		if err != nil {
+			return nil, 0, err
+		}
+
+		values := make([]int32, 0, len(strValues))
+		for _, value := range strValues {
+			intValue, err := strconv.Atoi(value)
+			if err != nil {
+				return nil, 0, errs.Wrap(err)
+			}
+			values = append(values, int32(intValue))
+		}
+
+		result[userID] = values
+	}
+
+	return result, nextCursor, nil
+}
+
 func (s *userOnline) SetUserOnline(ctx context.Context, userID string, online, offline []int32) error {
 	script := `
 	local key = KEYS[1]

pkg/localcache/lru/lru.go

@@ -20,7 +20,9 @@ type EvictCallback[K comparable, V any] simplelru.EvictCallback[K, V]
 
 type LRU[K comparable, V any] interface {
 	Get(key K, fetch func() (V, error)) (V, error)
+	Set(key K, value V)
 	SetHas(key K, value V) bool
+	GetBatch(keys []K, fetch func(keys []K) (map[K]V, error)) (map[K]V, error)
 	Del(key K) bool
 	Stop()
 }

pkg/localcache/lru/lru_expiration.go

@@ -51,6 +51,11 @@ type ExpirationLRU[K comparable, V any] struct {
 	target     Target
 }
 
+func (x *ExpirationLRU[K, V]) GetBatch(keys []K, fetch func(keys []K) (map[K]V, error)) (map[K]V, error) {
+	//TODO implement me
+	panic("implement me")
+}
+
 func (x *ExpirationLRU[K, V]) Get(key K, fetch func() (V, error)) (V, error) {
 	x.lock.Lock()
 	v, ok := x.core.Get(key)
@@ -99,5 +104,11 @@ func (x *ExpirationLRU[K, V]) SetHas(key K, value V) bool {
 	return false
 }
 
+func (x *ExpirationLRU[K, V]) Set(key K, value V) {
+	x.lock.Lock()
+	defer x.lock.Unlock()
+	x.core.Add(key, &expirationLruItem[V]{value: value})
+}
+
 func (x *ExpirationLRU[K, V]) Stop() {
 }

pkg/localcache/lru/lru_lazy.go

@@ -88,18 +88,76 @@ func (x *LayLRU[K, V]) Get(key K, fetch func() (V, error)) (V, error) {
 	return v.value, v.err
 }
 
-//func (x *LayLRU[K, V]) Set(key K, value V) {
-//	x.lock.Lock()
-//	x.core.Add(key, &layLruItem[V]{value: value, expires: time.Now().Add(x.successTTL).UnixMilli()})
-//	x.lock.Unlock()
-//}
-//
+func (x *LayLRU[K, V]) GetBatch(keys []K, fetch func(keys []K) (map[K]V, error)) (map[K]V, error) {
+	var (
+		err  error
+		once sync.Once
+	)
+
+	x.lock.Lock()
+	res := make(map[K]V)
+	queries := make([]K, 0)
+	setVs := make(map[K]*layLruItem[V])
+	for _, key := range keys {
+		v, ok := x.core.Get(key)
+		if ok {
+			x.lock.Unlock()
+			v.lock.Lock()
+			expires, value, err1 := v.expires, v.value, v.err
+			if expires != 0 && expires > time.Now().UnixMilli() {
+				v.lock.Unlock()
+				x.target.IncrGetHit()
+				res[key] = value
+				if err1 != nil {
+					once.Do(func() {
+						err = err1
+					})
+				}
+				continue
+			}
+		}
+		queries = append(queries, key)
+		x.lock.Unlock()
+	}
+	values, err1 := fetch(queries)
+	if err1 != nil {
+		once.Do(func() {
+			err = err1
+		})
+	}
+	for key, val := range values {
+		v := &layLruItem[V]{}
+		v.value = val
+
+		if err == nil {
+			v.expires = time.Now().Add(x.successTTL).UnixMilli()
+			x.target.IncrGetSuccess()
+		} else {
+			v.expires = time.Now().Add(x.failedTTL).UnixMilli()
+			x.target.IncrGetFailed()
+		}
+		setVs[key] = v
+		x.lock.Lock()
+		x.core.Add(key, v)
+		x.lock.Unlock()
+		res[key] = val
+	}
+
+	return res, err
+}
+
 //func (x *LayLRU[K, V]) Has(key K) bool {
 //	x.lock.Lock()
 //	defer x.lock.Unlock()
 //	return x.core.Contains(key)
 //}
 
+func (x *LayLRU[K, V]) Set(key K, value V) {
+	x.lock.Lock()
+	defer x.lock.Unlock()
+	x.core.Add(key, &layLruItem[V]{value: value, expires: time.Now().Add(x.successTTL).UnixMilli()})
+}
+
 func (x *LayLRU[K, V]) SetHas(key K, value V) bool {
 	x.lock.Lock()
 	defer x.lock.Unlock()

pkg/localcache/lru/lru_slot.go

@@ -32,6 +32,29 @@ type slotLRU[K comparable, V any] struct {
 	hash  func(k K) uint64
 }
 
+func (x *slotLRU[K, V]) GetBatch(keys []K, fetch func(keys []K) (map[K]V, error)) (map[K]V, error) {
+	var (
+		slotKeys = make(map[uint64][]K)
+		vs       = make(map[K]V)
+	)
+
+	for _, k := range keys {
+		index := x.getIndex(k)
+		slotKeys[index] = append(slotKeys[index], k)
+	}
+
+	for k, v := range slotKeys {
+		batches, err := x.slots[k].GetBatch(v, fetch)
+		if err != nil {
+			return nil, err
+		}
+		for key, value := range batches {
+			vs[key] = value
+		}
+	}
+	return vs, nil
+}
+
 func (x *slotLRU[K, V]) getIndex(k K) uint64 {
 	return x.hash(k) % x.n
 }
@@ -40,6 +63,10 @@ func (x *slotLRU[K, V]) Get(key K, fetch func() (V, error)) (V, error) {
 	return x.slots[x.getIndex(key)].Get(key, fetch)
 }
 
+func (x *slotLRU[K, V]) Set(key K, value V) {
+	x.slots[x.getIndex(key)].Set(key, value)
+}
+
 func (x *slotLRU[K, V]) SetHas(key K, value V) bool {
 	return x.slots[x.getIndex(key)].SetHas(key, value)
 }

pkg/rpccache/online.go

@@ -2,60 +2,197 @@ package rpccache
 
 import (
 	"context"
+	"fmt"
+	"github.com/openimsdk/protocol/constant"
+	"github.com/openimsdk/protocol/user"
+	"math/rand"
+	"strconv"
+	"sync"
+	"sync/atomic"
+	"time"
+
 	"github.com/openimsdk/open-im-server/v3/pkg/common/storage/cache/cachekey"
 	"github.com/openimsdk/open-im-server/v3/pkg/localcache"
 	"github.com/openimsdk/open-im-server/v3/pkg/localcache/lru"
 	"github.com/openimsdk/open-im-server/v3/pkg/rpcclient"
 	"github.com/openimsdk/open-im-server/v3/pkg/util/useronline"
+	"github.com/openimsdk/tools/db/cacheutil"
 	"github.com/openimsdk/tools/log"
 	"github.com/openimsdk/tools/mcontext"
 	"github.com/redis/go-redis/v9"
-	"math/rand"
-	"strconv"
-	"time"
 )
 
-func NewOnlineCache(user rpcclient.UserRpcClient, group *GroupLocalCache, rdb redis.UniversalClient, fn func(ctx context.Context, userID string, platformIDs []int32)) *OnlineCache {
+func NewOnlineCache(user rpcclient.UserRpcClient, group *GroupLocalCache, rdb redis.UniversalClient, fullUserCache bool, fn func(ctx context.Context, userID string, platformIDs []int32)) (*OnlineCache, error) {
+	l := &sync.Mutex{}
 	x := &OnlineCache{
-		user:  user,
-		group: group,
-		local: lru.NewSlotLRU(1024, localcache.LRUStringHash, func() lru.LRU[string, []int32] {
+		user:          user,
+		group:         group,
+		fullUserCache: fullUserCache,
+		Lock:          l,
+		Cond:          sync.NewCond(l),
+	}
+
+	ctx := mcontext.SetOperationID(context.TODO(), strconv.FormatInt(time.Now().UnixNano()+int64(rand.Uint32()), 10))
+
+	switch x.fullUserCache {
+	case true:
+		log.ZDebug(ctx, "fullUserCache is true")
+		x.mapCache = cacheutil.NewCache[string, []int32]()
+		go func() {
+			if err := x.initUsersOnlineStatus(ctx); err != nil {
+				log.ZError(ctx, "initUsersOnlineStatus failed", err)
+			}
+		}()
+	case false:
+		log.ZDebug(ctx, "fullUserCache is false")
+		x.lruCache = lru.NewSlotLRU(1024, localcache.LRUStringHash, func() lru.LRU[string, []int32] {
 			return lru.NewLayLRU[string, []int32](2048, cachekey.OnlineExpire/2, time.Second*3, localcache.EmptyTarget{}, func(key string, value []int32) {})
-		}),
+		})
+		x.CurrentPhase.Store(DoSubscribeOver)
+		x.Cond.Broadcast()
 	}
+
 	go func() {
-		ctx := mcontext.SetOperationID(context.Background(), cachekey.OnlineChannel+strconv.FormatUint(rand.Uint64(), 10))
-		for message := range rdb.Subscribe(ctx, cachekey.OnlineChannel).Channel() {
-			userID, platformIDs, err := useronline.ParseUserOnlineStatus(message.Payload)
+		x.doSubscribe(ctx, rdb, fn)
+	}()
+	return x, nil
+}
+
+const (
+	Begin uint32 = iota
+	DoOnlineStatusOver
+	DoSubscribeOver
+)
+
+type OnlineCache struct {
+	user  rpcclient.UserRpcClient
+	group *GroupLocalCache
+
+	// fullUserCache if enabled, caches the online status of all users using mapCache;
+	// otherwise, only a portion of users' online statuses (regardless of whether they are online) will be cached using lruCache.
+	fullUserCache bool
+
+	lruCache lru.LRU[string, []int32]
+	mapCache *cacheutil.Cache[string, []int32]
+
+	Lock         *sync.Mutex
+	Cond         *sync.Cond
+	CurrentPhase atomic.Uint32
+}
+
+func (o *OnlineCache) initUsersOnlineStatus(ctx context.Context) (err error) {
+	log.ZDebug(ctx, "init users online status begin")
+
+	var (
+		totalSet      atomic.Int64
+		maxTries      = 5
+		retryInterval = time.Second * 5
+
+		resp *user.GetAllOnlineUsersResp
+	)
+
+	defer func(t time.Time) {
+		log.ZInfo(ctx, "init users online status end", "cost", time.Since(t), "totalSet", totalSet.Load())
+		o.CurrentPhase.Store(DoOnlineStatusOver)
+		o.Cond.Broadcast()
+	}(time.Now())
+
+	retryOperation := func(operation func() error, operationName string) error {
+		for i := 0; i < maxTries; i++ {
+			if err = operation(); err != nil {
+				log.ZWarn(ctx, fmt.Sprintf("initUsersOnlineStatus: %s failed", operationName), err)
+				time.Sleep(retryInterval)
+			} else {
+				return nil
+			}
+		}
+		return err
+	}
+
+	cursor := uint64(0)
+	for resp == nil || resp.NextCursor != 0 {
+		if err = retryOperation(func() error {
+			resp, err = o.user.GetAllOnlineUsers(ctx, cursor)
 			if err != nil {
-				log.ZError(ctx, "OnlineCache setUserOnline redis subscribe parseUserOnlineStatus", err, "payload", message.Payload, "channel", message.Channel)
-				continue
+				return err
+			}
+
+			for _, u := range resp.StatusList {
+				if u.Status == constant.Online {
+					o.setUserOnline(u.UserID, u.PlatformIDs)
+				}
+				totalSet.Add(1)
 			}
-			storageCache := x.setUserOnline(userID, platformIDs)
-			log.ZDebug(ctx, "OnlineCache setUserOnline", "userID", userID, "platformIDs", platformIDs, "payload", message.Payload, "storageCache", storageCache)
+			cursor = resp.NextCursor
+			return nil
+		}, "getAllOnlineUsers"); err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+func (o *OnlineCache) doSubscribe(ctx context.Context, rdb redis.UniversalClient, fn func(ctx context.Context, userID string, platformIDs []int32)) {
+	o.Lock.Lock()
+	ch := rdb.Subscribe(ctx, cachekey.OnlineChannel).Channel()
+	for o.CurrentPhase.Load() < DoOnlineStatusOver {
+		o.Cond.Wait()
+	}
+	o.Lock.Unlock()
+	log.ZInfo(ctx, "begin doSubscribe")
+
+	doMessage := func(message *redis.Message) {
+		userID, platformIDs, err := useronline.ParseUserOnlineStatus(message.Payload)
+		if err != nil {
+			log.ZError(ctx, "OnlineCache setHasUserOnline redis subscribe parseUserOnlineStatus", err, "payload", message.Payload, "channel", message.Channel)
+			return
+		}
+		log.ZDebug(ctx, fmt.Sprintf("get subscribe %s message", cachekey.OnlineChannel), "useID", userID, "platformIDs", platformIDs)
+		switch o.fullUserCache {
+		case true:
+			if len(platformIDs) == 0 {
+				// offline
+				o.mapCache.Delete(userID)
+			} else {
+				o.mapCache.Store(userID, platformIDs)
+			}
+		case false:
+			storageCache := o.setHasUserOnline(userID, platformIDs)
+			log.ZDebug(ctx, "OnlineCache setHasUserOnline", "userID", userID, "platformIDs", platformIDs, "payload", message.Payload, "storageCache", storageCache)
 			if fn != nil {
 				fn(ctx, userID, platformIDs)
 			}
 		}
-	}()
-	return x
-}
+	}
 
-type OnlineCache struct {
-	user  rpcclient.UserRpcClient
-	group *GroupLocalCache
-	local lru.LRU[string, []int32]
+	if o.CurrentPhase.Load() == DoOnlineStatusOver {
+		for done := false; !done; {
+			select {
+			case message := <-ch:
+				doMessage(message)
+			default:
+				o.CurrentPhase.Store(DoSubscribeOver)
+				o.Cond.Broadcast()
+				done = true
+			}
+		}
+	}
+
+	for message := range ch {
+		doMessage(message)
+	}
 }
 
 func (o *OnlineCache) getUserOnlinePlatform(ctx context.Context, userID string) ([]int32, error) {
-	platformIDs, err := o.local.Get(userID, func() ([]int32, error) {
+	platformIDs, err := o.lruCache.Get(userID, func() ([]int32, error) {
 		return o.user.GetUserOnlinePlatform(ctx, userID)
 	})
 	if err != nil {
 		log.ZError(ctx, "OnlineCache GetUserOnlinePlatform", err, "userID", userID)
 		return nil, err
 	}
-	log.ZDebug(ctx, "OnlineCache GetUserOnlinePlatform", "userID", userID, "platformIDs", platformIDs)
+	//log.ZDebug(ctx, "OnlineCache GetUserOnlinePlatform", "userID", userID, "platformIDs", platformIDs)
 	return platformIDs, nil
 }
 
@@ -69,6 +206,16 @@ func (o *OnlineCache) GetUserOnlinePlatform(ctx context.Context, userID string)
 	return platformIDs, nil
 }
 
+// func (o *OnlineCache) GetUserOnlinePlatformBatch(ctx context.Context, userIDs []string) (map[string]int32, error) {
+// 	platformIDs, err := o.getUserOnlinePlatform(ctx, userIDs)
+// 	if err != nil {
+// 		return nil, err
+// 	}
+// 	tmp := make([]int32, len(platformIDs))
+// 	copy(tmp, platformIDs)
+// 	return platformIDs, nil
+// }
+
 func (o *OnlineCache) GetUserOnline(ctx context.Context, userID string) (bool, error) {
 	platformIDs, err := o.getUserOnlinePlatform(ctx, userID)
 	if err != nil {
@@ -77,10 +224,68 @@ func (o *OnlineCache) GetUserOnline(ctx context.Context, userID string) (bool, e
 	return len(platformIDs) > 0, nil
 }
 
+func (o *OnlineCache) getUserOnlinePlatformBatch(ctx context.Context, userIDs []string) (map[string][]int32, error) {
+	platformIDsMap, err := o.lruCache.GetBatch(userIDs, func(missingUsers []string) (map[string][]int32, error) {
+		platformIDsMap := make(map[string][]int32)
+
+		usersStatus, err := o.user.GetUsersOnlinePlatform(ctx, missingUsers)
+		if err != nil {
+			return nil, err
+		}
+
+		for _, u := range usersStatus {
+			platformIDsMap[u.UserID] = u.PlatformIDs
+		}
+
+		return platformIDsMap, nil
+	})
+	if err != nil {
+		log.ZError(ctx, "OnlineCache GetUserOnlinePlatform", err, "userID", userIDs)
+		return nil, err
+	}
+	return platformIDsMap, nil
+}
+
+func (o *OnlineCache) GetUsersOnline(ctx context.Context, userIDs []string) ([]string, []string, error) {
+	t := time.Now()
+
+	var (
+		onlineUserIDs  = make([]string, 0, len(userIDs))
+		offlineUserIDs = make([]string, 0, len(userIDs))
+	)
+
+	switch o.fullUserCache {
+	case true:
+		for _, userID := range userIDs {
+			if _, ok := o.mapCache.Load(userID); ok {
+				onlineUserIDs = append(onlineUserIDs, userID)
+			} else {
+				offlineUserIDs = append(offlineUserIDs, userID)
+			}
+		}
+	case false:
+		userOnlineMap, err := o.getUserOnlinePlatformBatch(ctx, userIDs)
+		if err != nil {
+			return nil, nil, err
+		}
+
+		for key, value := range userOnlineMap {
+			if len(value) > 0 {
+				onlineUserIDs = append(onlineUserIDs, key)
+			} else {
+				offlineUserIDs = append(offlineUserIDs, key)
+			}
+		}
+	}
+
+	log.ZInfo(ctx, "get users online", "online users length", len(userIDs), "offline users length", len(offlineUserIDs), "cost", time.Since(t))
+	return userIDs, offlineUserIDs, nil
+}
+
 //func (o *OnlineCache) GetUsersOnline(ctx context.Context, userIDs []string) ([]string, error) {
 //	onlineUserIDs := make([]string, 0, len(userIDs))
 //	for _, userID := range userIDs {
-//		online, err := o.GetUserOnline(ctx, userID)
+// online, err := o.GetUserOnline(ctx, userID)
 //		if err != nil {
 //			return nil, err
 //		}
@@ -111,6 +316,15 @@ func (o *OnlineCache) GetUserOnline(ctx context.Context, userID string) (bool, e
 //	return onlineUserIDs, nil
 //}
 
-func (o *OnlineCache) setUserOnline(userID string, platformIDs []int32) bool {
-	return o.local.SetHas(userID, platformIDs)
+func (o *OnlineCache) setUserOnline(userID string, platformIDs []int32) {
+	switch o.fullUserCache {
+	case true:
+		o.mapCache.Store(userID, platformIDs)
+	case false:
+		o.lruCache.Set(userID, platformIDs)
+	}
+}
+
+func (o *OnlineCache) setHasUserOnline(userID string, platformIDs []int32) bool {
+	return o.lruCache.SetHas(userID, platformIDs)
 }

pkg/rpcclient/user.go

@@ -169,6 +169,15 @@ func (u *UserRpcClient) Access(ctx context.Context, ownerUserID string) error {
 	return authverify.CheckAccessV3(ctx, ownerUserID, u.imAdminUserID)
 }
 
+// GetAllUserID retrieves all user IDs with pagination options.
+func (u *UserRpcClient) GetAllUserID(ctx context.Context, pageNumber, showNumber int32) (*user.GetAllUserIDResp, error) {
+	resp, err := u.Client.GetAllUserID(ctx, &user.GetAllUserIDReq{Pagination: &sdkws.RequestPagination{PageNumber: pageNumber, ShowNumber: showNumber}})
+	if err != nil {
+		return nil, err
+	}
+	return resp, nil
+}
+
 // GetAllUserIDs retrieves all user IDs with pagination options.
 func (u *UserRpcClient) GetAllUserIDs(ctx context.Context, pageNumber, showNumber int32) ([]string, error) {
 	resp, err := u.Client.GetAllUserID(ctx, &user.GetAllUserIDReq{Pagination: &sdkws.RequestPagination{PageNumber: pageNumber, ShowNumber: showNumber}})
@@ -215,3 +224,7 @@ func (u *UserRpcClient) GetUserOnlinePlatform(ctx context.Context, userID string
 	}
 	return resp[0].PlatformIDs, nil
 }
+
+func (u *UserRpcClient) GetAllOnlineUsers(ctx context.Context, cursor uint64) (*user.GetAllOnlineUsersResp, error) {
+	return u.Client.GetAllOnlineUsers(ctx, &user.GetAllOnlineUsersReq{Cursor: cursor})
+}

start-config.yml

@@ -3,8 +3,8 @@ serviceBinaries:
   openim-crontask: 1
   openim-rpc-user: 1
   openim-msggateway: 1
-  openim-push: 4
-  openim-msgtransfer: 4
+  openim-push: 8
+  openim-msgtransfer: 8
   openim-rpc-conversation: 1
   openim-rpc-auth: 1
   openim-rpc-group: 1