ebpf的主从同步实现，QPS测试与内存池QPS测试。

2026-01-30 16:00:06 +00:00
parent 2bdb48d63d
commit fbdcff6878
23 changed files with 599 additions and 383 deletions
--- a/8
+++ b/8
@@ -12,7 +12,8 @@ SRCS 		=  	$(NET_SRCS) $(KV_SRCS) $(MEM_SRCS) $(COMMON_SRCS) $(DUMP_SRCS)

 INC = -I./NtyCo/core/ -I/usr/include/libxml2 -I./
 LIBDIRS = -L./NtyCo/ 
-LIBS = -lntyco -lpthread -luring -ldl -lxml2 -ljemalloc
+# LIBS = -lntyco -lpthread -luring -ldl -lxml2 -ljemalloc
+LIBS = -lntyco -lpthread -luring -ldl -lxml2

 TARGET = kvstore
 SUBDIR = ./NtyCo/
@@ -42,8 +43,11 @@ $(TEST_REDIS): $(TEST_REDIS_SRCS)
 %.o: %.c
 	$(CC) $(CFLAGS) $(INC) -c $^ -g -o $@

-clean: clmem
+clean: clmem cldata
 	rm -rf $(OBJS) $(TARGET) $(TEST_REDIS)

 clmem:
 	rm -rf mem_leak/*
+
+cldata:
+	rm -rf data/*
--- a/README.md
+++ b/README.md
@@ -3,14 +3,14 @@
 ## 需求
 1. ntyco需要作为kvstore的submodule,通过git clone一次下载。 **完成**。
 2. README需要包含编译步骤，测试方案与可行性，性能数据。  **完成**。
-3. 全量持久化保存数据集。 **BUG FIX，完成**。
+3. 全量持久化保存数据集。 **BUG FIX**。
 4. 持久化的性能数据。
 5. 特殊字符，可以解决redis的resp协议。 **完成**。
 6. 实现配置文件，把日志级别，端口ip，主从模式，持久化方案。 **完成**。
 7. 持久化落盘用io_uring，加载配置文件用mmap。   **完成**。
-8. 主从同步的性能，开启与关闭性能做到。
-9. 主从同步600w条,出现的coredump。
-10. 主从同步用ebpf实现。
+8. 主从同步的性能，开启与关闭性能做到5%?。
+9. 主从同步600w条,出现的coredump。     **完成**。
+10. 主从同步用ebpf实现。               **BUG FIX**。
 11. 内存池测试qps与虚拟内存，物理内存。
 12. 实现一个内存泄露检测组件。          **完成**。

@@ -34,6 +34,191 @@ make
 ```

 ## 测试
+### 测试1：性能测试
+测试条件：
+1. 不启用持久化。
+2. 不启用主从同步。
+2. pipline：
+    1. RSET 100w 条， p:i v:i -> +OK 
+    2. RGET 100w 条， p:i -> +v:i
+    3. RDEL 100w 条。 p:i -> +OK
+3. 重复 15 次.
+4. 本机发送请求。
+
+内存分配: malloc
+```bash
+lian@ubuntu:~/share/9.1-kvstore$ ./test-redis/testcase 192.168.10.129 8888 3
+Connected to 192.168.10.129:8888
+BATCH (N=3000000) --> time_used=3294 ms, qps=910746
+BATCH (N=3000000) --> time_used=3501 ms, qps=856898
+BATCH (N=3000000) --> time_used=3457 ms, qps=867804
+BATCH (N=3000000) --> time_used=3351 ms, qps=895255
+BATCH (N=3000000) --> time_used=3320 ms, qps=903614
+BATCH (N=3000000) --> time_used=3551 ms, qps=844832
+BATCH (N=3000000) --> time_used=3354 ms, qps=894454
+BATCH (N=3000000) --> time_used=3475 ms, qps=863309
+BATCH (N=3000000) --> time_used=3404 ms, qps=881316
+BATCH (N=3000000) --> time_used=3460 ms, qps=867052
+BATCH (N=3000000) --> time_used=3392 ms, qps=884433
+BATCH (N=3000000) --> time_used=3427 ms, qps=875401
+BATCH (N=3000000) --> time_used=3441 ms, qps=871839
+BATCH (N=3000000) --> time_used=3471 ms, qps=864304
+BATCH (N=3000000) --> time_used=3354 ms, qps=894454
+BATCH (N=3000000) --> time_used=3447 ms, qps=870322
+BATCH (N=3000000) --> time_used=3364 ms, qps=891795
+BATCH (N=3000000) --> time_used=3200 ms, qps=937500
+BATCH (N=3000000) --> time_used=3159 ms, qps=949667
+BATCH (N=3000000) --> time_used=3482 ms, qps=861573
+BATCH (N=3000000) --> time_used=3474 ms, qps=863557
+BATCH (N=3000000) --> time_used=3591 ms, qps=835421
+BATCH (N=3000000) --> time_used=3466 ms, qps=865551
+BATCH (N=3000000) --> time_used=3425 ms, qps=875912
+BATCH (N=3000000) --> time_used=3346 ms, qps=896592
+BATCH (N=3000000) --> time_used=3532 ms, qps=849377
+BATCH (N=3000000) --> time_used=3471 ms, qps=864304
+BATCH (N=3000000) --> time_used=3616 ms, qps=829646
+BATCH (N=3000000) --> time_used=3403 ms, qps=881575
+BATCH (N=3000000) --> time_used=3426 ms, qps=875656
+BATCH (N=3000000) --> time_used=3493 ms, qps=858860
+BATCH (N=3000000) --> time_used=3411 ms, qps=879507
+BATCH (N=3000000) --> time_used=3422 ms, qps=876680
+BATCH (N=3000000) --> time_used=3556 ms, qps=843644
+BATCH (N=3000000) --> time_used=3285 ms, qps=913242
+BATCH (N=3000000) --> time_used=3486 ms, qps=860585
+BATCH (N=3000000) --> time_used=3427 ms, qps=875401
+BATCH (N=3000000) --> time_used=3563 ms, qps=841987
+BATCH (N=3000000) --> time_used=3304 ms, qps=907990
+BATCH (N=3000000) --> time_used=3582 ms, qps=837520
+BATCH (N=3000000) --> time_used=3468 ms, qps=865051
+BATCH (N=3000000) --> time_used=3360 ms, qps=892857
+BATCH (N=3000000) --> time_used=3426 ms, qps=875656
+BATCH (N=3000000) --> time_used=3186 ms, qps=941619
+BATCH (N=3000000) --> time_used=3251 ms, qps=922792
+BATCH (N=3000000) --> time_used=3400 ms, qps=882352
+BATCH (N=3000000) --> time_used=3446 ms, qps=870574
+BATCH (N=3000000) --> time_used=3302 ms, qps=908540
+BATCH (N=3000000) --> time_used=3072 ms, qps=976562
+BATCH (N=3000000) --> time_used=3458 ms, qps=867553
+average qps:880462
+ALL TESTS PASSED.
+```
+
+内存分配: 自实现内存池
+```bash
+lian@ubuntu:~/share/9.1-kvstore$ ./test-redis/testcase 192.168.10.129 8888 3
+Connected to 192.168.10.129:8888
+BATCH (N=3000000) --> time_used=3552 ms, qps=844594
+BATCH (N=3000000) --> time_used=3404 ms, qps=881316
+BATCH (N=3000000) --> time_used=3083 ms, qps=973078
+BATCH (N=3000000) --> time_used=3315 ms, qps=904977
+BATCH (N=3000000) --> time_used=3462 ms, qps=866551
+BATCH (N=3000000) --> time_used=3334 ms, qps=899820
+BATCH (N=3000000) --> time_used=3358 ms, qps=893388
+BATCH (N=3000000) --> time_used=3423 ms, qps=876424
+BATCH (N=3000000) --> time_used=3319 ms, qps=903886
+BATCH (N=3000000) --> time_used=3327 ms, qps=901713
+BATCH (N=3000000) --> time_used=3157 ms, qps=950269
+BATCH (N=3000000) --> time_used=3241 ms, qps=925640
+BATCH (N=3000000) --> time_used=3301 ms, qps=908815
+BATCH (N=3000000) --> time_used=3345 ms, qps=896860
+BATCH (N=3000000) --> time_used=3319 ms, qps=903886
+BATCH (N=3000000) --> time_used=3312 ms, qps=905797
+BATCH (N=3000000) --> time_used=3337 ms, qps=899011
+BATCH (N=3000000) --> time_used=3309 ms, qps=906618
+BATCH (N=3000000) --> time_used=3385 ms, qps=886262
+BATCH (N=3000000) --> time_used=3328 ms, qps=901442
+BATCH (N=3000000) --> time_used=3194 ms, qps=939261
+BATCH (N=3000000) --> time_used=3309 ms, qps=906618
+BATCH (N=3000000) --> time_used=3262 ms, qps=919681
+BATCH (N=3000000) --> time_used=3314 ms, qps=905250
+BATCH (N=3000000) --> time_used=3382 ms, qps=887049
+BATCH (N=3000000) --> time_used=3296 ms, qps=910194
+BATCH (N=3000000) --> time_used=3331 ms, qps=900630
+BATCH (N=3000000) --> time_used=3279 ms, qps=914913
+BATCH (N=3000000) --> time_used=2996 ms, qps=1001335
+BATCH (N=3000000) --> time_used=3387 ms, qps=885739
+BATCH (N=3000000) --> time_used=3346 ms, qps=896592
+BATCH (N=3000000) --> time_used=3241 ms, qps=925640
+BATCH (N=3000000) --> time_used=3353 ms, qps=894721
+BATCH (N=3000000) --> time_used=3366 ms, qps=891265
+BATCH (N=3000000) --> time_used=3387 ms, qps=885739
+BATCH (N=3000000) --> time_used=3382 ms, qps=887049
+BATCH (N=3000000) --> time_used=3358 ms, qps=893388
+BATCH (N=3000000) --> time_used=3372 ms, qps=889679
+BATCH (N=3000000) --> time_used=3467 ms, qps=865301
+BATCH (N=3000000) --> time_used=3243 ms, qps=925069
+BATCH (N=3000000) --> time_used=3191 ms, qps=940144
+BATCH (N=3000000) --> time_used=3365 ms, qps=891530
+BATCH (N=3000000) --> time_used=3218 ms, qps=932256
+BATCH (N=3000000) --> time_used=3332 ms, qps=900360
+BATCH (N=3000000) --> time_used=3172 ms, qps=945775
+BATCH (N=3000000) --> time_used=3211 ms, qps=934288
+BATCH (N=3000000) --> time_used=3124 ms, qps=960307
+BATCH (N=3000000) --> time_used=3043 ms, qps=985869
+BATCH (N=3000000) --> time_used=3086 ms, qps=972132
+BATCH (N=3000000) --> time_used=3201 ms, qps=937207
+average qps:911106
+ALL TESTS PASSED.
+```
+
+内存分配：jemalloc
+```shell
+lian@ubuntu:~/share/9.1-kvstore$ ./test-redis/testcase 192.168.10.129 8888 3
+Connected to 192.168.10.129:8888
+BATCH (N=3000000) --> time_used=3197 ms, qps=938379
+BATCH (N=3000000) --> time_used=3221 ms, qps=931387
+BATCH (N=3000000) --> time_used=3360 ms, qps=892857
+BATCH (N=3000000) --> time_used=3292 ms, qps=911300
+BATCH (N=3000000) --> time_used=3407 ms, qps=880540
+BATCH (N=3000000) --> time_used=3317 ms, qps=904431
+BATCH (N=3000000) --> time_used=3337 ms, qps=899011
+BATCH (N=3000000) --> time_used=3384 ms, qps=886524
+BATCH (N=3000000) --> time_used=3355 ms, qps=894187
+BATCH (N=3000000) --> time_used=3379 ms, qps=887836
+BATCH (N=3000000) --> time_used=3243 ms, qps=925069
+BATCH (N=3000000) --> time_used=3377 ms, qps=888362
+BATCH (N=3000000) --> time_used=3212 ms, qps=933997
+BATCH (N=3000000) --> time_used=3248 ms, qps=923645
+BATCH (N=3000000) --> time_used=3234 ms, qps=927643
+BATCH (N=3000000) --> time_used=3152 ms, qps=951776
+BATCH (N=3000000) --> time_used=3089 ms, qps=971188
+BATCH (N=3000000) --> time_used=3287 ms, qps=912686
+BATCH (N=3000000) --> time_used=3079 ms, qps=974342
+BATCH (N=3000000) --> time_used=3261 ms, qps=919963
+BATCH (N=3000000) --> time_used=3123 ms, qps=960614
+BATCH (N=3000000) --> time_used=3234 ms, qps=927643
+BATCH (N=3000000) --> time_used=3056 ms, qps=981675
+BATCH (N=3000000) --> time_used=3040 ms, qps=986842
+BATCH (N=3000000) --> time_used=3187 ms, qps=941324
+BATCH (N=3000000) --> time_used=3311 ms, qps=906070
+BATCH (N=3000000) --> time_used=3155 ms, qps=950871
+BATCH (N=3000000) --> time_used=3318 ms, qps=904159
+BATCH (N=3000000) --> time_used=3372 ms, qps=889679
+BATCH (N=3000000) --> time_used=3254 ms, qps=921942
+BATCH (N=3000000) --> time_used=3386 ms, qps=886001
+BATCH (N=3000000) --> time_used=3413 ms, qps=878992
+BATCH (N=3000000) --> time_used=3474 ms, qps=863557
+BATCH (N=3000000) --> time_used=3412 ms, qps=879249
+BATCH (N=3000000) --> time_used=3414 ms, qps=878734
+BATCH (N=3000000) --> time_used=3325 ms, qps=902255
+BATCH (N=3000000) --> time_used=3346 ms, qps=896592
+BATCH (N=3000000) --> time_used=3345 ms, qps=896860
+BATCH (N=3000000) --> time_used=3582 ms, qps=837520
+BATCH (N=3000000) --> time_used=3412 ms, qps=879249
+BATCH (N=3000000) --> time_used=3370 ms, qps=890207
+BATCH (N=3000000) --> time_used=3375 ms, qps=888888
+BATCH (N=3000000) --> time_used=3190 ms, qps=940438
+BATCH (N=3000000) --> time_used=3324 ms, qps=902527
+BATCH (N=3000000) --> time_used=3253 ms, qps=922225
+BATCH (N=3000000) --> time_used=3230 ms, qps=928792
+BATCH (N=3000000) --> time_used=3294 ms, qps=910746
+BATCH (N=3000000) --> time_used=3295 ms, qps=910470
+BATCH (N=3000000) --> time_used=3148 ms, qps=952986
+BATCH (N=3000000) --> time_used=3228 ms, qps=929368
+average qps:914031
+ALL TESTS PASSED.
+```
+

 ### 面试题
 1. 为什么会实现kvstore，使用场景在哪里？
--- a/config/config.xml
+++ b/config/config.xml
@@ -17,7 +17,7 @@
    </log>

    <persistence>
-        <type>incremental</type> <!-- incremental / none -->
+        <type>none</type> <!-- incremental / none -->
        <dir>data</dir>          <!-- 所有持久化文件所在目录 -->
        
        <wal>kvs_oplog.db</wal>
@@ -27,7 +27,7 @@
    </persistence>

    <memory>
-        <allocator>mypool</allocator> <!-- jemalloc / malloc / mypool -->
+        <allocator>malloc</allocator> <!-- jemalloc / malloc / mypool -->
        <leakage>disable</leakage>       <!-- enable/disable-->
    </memory>
 </config>
--- a/diskuring/diskuring.c
+++ b/diskuring/diskuring.c
@@ -110,6 +110,13 @@ static void *worker_main(void *arg)
            task_t *done = (task_t *)(uintptr_t)cqe->user_data;
            task_finish(done, cqe->res);

+            // todo: 失败应该通知主线程，提供一个链表，把失败的taskpush进去。主线程用定时器任务定时处理失败信息然后destory。
+            // 暂时用打印说明失败
+            if(done->res < 0){
+                printf("uring failed: fd:%d, offset:%ld\n", done->fd, done->off);
+            }
+            task_destroy(done);
+
            io_uring_cqe_seen(&ctx->ring, cqe);
        }
    }
--- a/dump/kvs_dump.h
+++ b/dump/kvs_dump.h
@@ -11,7 +11,6 @@ extern char global_hash_file[256];

 int kvs_create_snapshot(iouring_ctx_t *uring, const char* array_file, const char* rbtree_file, const char* hash_file);
 int kvs_create_snapshot_async(const char *ip, int port);
-void __complete_snapshot(const char *ip, int port, const char *array_file, const char *rbtree_file, const char *hash_file);

 extern int global_oplog_fd;

--- a/dump/kvs_oplog.c
+++ b/dump/kvs_oplog.c
@@ -54,8 +54,8 @@ int kvs_oplog_append(const uint8_t *cmd, size_t len, int logfd){
        return -4;
    }

-    task_wait(t);
-    task_destroy(t);
+    // task_wait(t);
+    // task_destroy(t);

 	return 0;
 }
--- a/dump/kvs_snapshot.c
+++ b/dump/kvs_snapshot.c
@@ -1,5 +1,6 @@
 #include "kvstore.h"
 #include "diskuring/diskuring.h"
+#include "kvs_dump.h"
 #include <unistd.h>
 #include <arpa/inet.h>
 #include <sys/socket.h>
@@ -39,10 +40,6 @@ int kvs_create_snapshot(iouring_ctx_t *uring, const char* array_file, const char
 	return ret;
 }

-void __complete_snapshot(const char *ip, int port, const char *array_file, const char *rbtree_file, const char *hash_file){
-	
-}
-
 static int send_file_to_ipport(const char *ip, int port, const char *filename) {
 	int sockfd = socket(AF_INET, SOCK_STREAM, 0);
 	if (sockfd < 0) { perror("socket"); return -1; }
@@ -119,9 +116,6 @@ int kvs_create_snapshot_async(const char *ip, int port){
            _exit(1);
        }

-		// hook 通知 eBPF
-		__complete_snapshot(ip, port, tmp_array, tmp_rbtree, tmp_hash);
-
 		iouring_shutdown(&uring);
 		_exit(0);
 	} else {
--- a/ebpf/c/replica.bpf.c
+++ b/ebpf/c/replica.bpf.c
@@ -1,103 +1,80 @@
 // SPDX-License-Identifier: GPL-2.0 OR BSD-3-Clause
 /* Copyright (c) 2020 Facebook */
-#include <linux/bpf.h>
-#include <linux/ptrace.h>
+#include "vmlinux.h"
 #include <bpf/bpf_helpers.h>
 #include <bpf/bpf_tracing.h>
+#include <bpf/bpf_core_read.h>

 #include "replica.h"

 char LICENSE[] SEC("license") = "Dual BSD/GPL";

-int my_pid = 0;
-
 struct {
-	__uint(type, BPF_MAP_TYPE_PERF_EVENT_ARRAY);
-	__uint(key_size, sizeof(int));
-	__uint(value_size, sizeof(int));
-} channel SEC(".maps");
+    __uint(type, BPF_MAP_TYPE_PERF_EVENT_ARRAY);
+    __uint(key_size, sizeof(int));
+    __uint(value_size, sizeof(int));
+} events SEC(".maps");

-SEC("uprobe/kvs_create_snapshot_async")
-int uprobe_create_snapshot_async(struct pt_regs *ctx)
+/* __completed_cmd(const uint8_t *cmd, size_t len, unsigned long long seq); */
+SEC("uprobe//home/lian/share/9.1-kvstore/kvstore:__completed_cmd")
+int BPF_KPROBE(handle_completed_cmd,
+               const __u8 *cmd, size_t len, __u64 seq)
 {
-    struct event ev;
-	__builtin_memset(&ev, 0, sizeof(ev));
+    struct replica_event evt = {};
+    __u32 copy_len;

-    const char *ip;
-    __u32 port;
+    evt.type = EVENT_COMPLETED_CMD;
+    evt.complete.seq = seq;

-    ev.type = EVENT_CREATE_SNAPSHOT_ASYNC;
+    copy_len = len;
+    if (copy_len > MAX_CMD_LEN)
+        copy_len = MAX_CMD_LEN;

-    ip   = (const char *)PT_REGS_PARM1(ctx);
-    port = (__u32)PT_REGS_PARM2(ctx);
+    evt.complete.len = copy_len;

-    bpf_probe_read_user_str(ev.data.sync.ip,
-                            sizeof(ev.data.sync.ip),
-                            ip);
-    ev.data.sync.port = port;
+    if (cmd)
+        bpf_probe_read_user(evt.complete.cmd, copy_len, cmd);

-    bpf_perf_event_output(ctx, &channel,
+    bpf_perf_event_output(ctx, &events,
                          BPF_F_CURRENT_CPU,
-                          &ev, sizeof(ev));
+                          &evt, sizeof(evt));
    return 0;
 }

-SEC("uprobe/__compeleted_cmd")
-int uprobe_completed_cmd(struct pt_regs *ctx)
+/* __ssync(const uint8_t *ip, uint32_t ip_len, int port, unsigned long long seq); */
+SEC("uprobe//home/lian/share/9.1-kvstore/kvstore:__ssync")
+int BPF_KPROBE(handle_ssync,
+               const __u8 *ip, __u32 ip_len, int port, __u64 seq)
 {
-    struct event ev;
-	__builtin_memset(&ev, 0, sizeof(ev));
+    struct replica_event evt = {};

-    const __u8 *cmd;
-    __u32 len;
+    evt.type = EVENT_SSYNC;
+    evt.sync.seq = seq;
+    evt.sync.port = port;

-    ev.type = EVENT_COMPLETED_CMD;
+    __u32 copy_len = ip_len;
+    if (copy_len > sizeof(evt.sync.ip))
+        copy_len = sizeof(evt.sync.ip);

-    cmd = (const __u8 *)PT_REGS_PARM1(ctx);
-    len = (__u32)PT_REGS_PARM2(ctx);
+    if (ip)
+        bpf_probe_read_user(evt.sync.ip, copy_len, ip);

-    if (len > sizeof(ev.data.cmd.cmd))
-        len = sizeof(ev.data.cmd.cmd);
-
-    ev.data.cmd.len = len;
-
-    bpf_probe_read_user(ev.data.cmd.cmd, len, cmd);
-
-    bpf_perf_event_output(ctx, &channel,
+    bpf_perf_event_output(ctx, &events,
                          BPF_F_CURRENT_CPU,
-                          &ev, sizeof(ev));
+                          &evt, sizeof(evt));
    return 0;
 }

-
-SEC("uprobe/__create_snapshot_ok")
-int uprobe_create_snapshot_ok(struct pt_regs *ctx)
+/* __sready(void); */
+SEC("uprobe//home/lian/share/9.1-kvstore/kvstore:__sready")
+int BPF_KPROBE(handle_sready)
 {
-    struct event ev;
-	__builtin_memset(&ev, 0, sizeof(ev));
+    struct replica_event evt = {};

-    const char *array_file;
-    const char *rbtree_file;
-    const char *hash_file;
+    evt.type = EVENT_SREADY;

-    ev.type = EVENT_CREATE_SNAPSHOT_OK;
-
-    array_file  = (const char *)PT_REGS_PARM1(ctx);
-    rbtree_file = (const char *)PT_REGS_PARM2(ctx);
-    hash_file   = (const char *)PT_REGS_PARM3(ctx);
-
-    bpf_probe_read_user_str(ev.data.ok.array_file,
-                            sizeof(ev.data.ok.array_file),
-                            array_file);
-    bpf_probe_read_user_str(ev.data.ok.rbtree_file,
-                            sizeof(ev.data.ok.rbtree_file),
-                            rbtree_file);
-    bpf_probe_read_user_str(ev.data.ok.hash_file,
-                            sizeof(ev.data.ok.hash_file),
-                            hash_file);
-
-    bpf_perf_event_output(ctx, &channel,
+    bpf_perf_event_output(ctx, &events,
                          BPF_F_CURRENT_CPU,
-                          &ev, sizeof(ev));
+                          &evt, sizeof(evt));
    return 0;
 }
--- a/ebpf/c/replica.c
+++ b/ebpf/c/replica.c
@@ -2,6 +2,8 @@
 /* Copyright (c) 2020 Facebook */
 #include <stdio.h>
 #include <unistd.h>
+#include <stdlib.h>
+#include <string.h>
 #include <sys/resource.h>
 #include <bpf/libbpf.h>
 #include "replica.skel.h"
@@ -12,10 +14,15 @@

 #include "replica.h"

+typedef enum {
+    OFFLINE = 0,
+    ONLINE  = 1,
+}replica_state_e ;

 struct cmd_node {
+    __u64 seq;
+    __u32 len;
    uint8_t *cmd;
-    size_t len;
    struct cmd_node *next;
 };

@@ -25,164 +32,184 @@ struct pending_queue {
 	int count;
 };

-static void queue_init(struct pending_queue *q) {
-    q->head = q->tail = NULL;
-	q->count = 0;
-}
+/* ================= 全局状态 ================= */

-static void queue_push(struct pending_queue *q, const uint8_t *cmd, size_t len) {
-    struct cmd_node *node = malloc(sizeof(*node));
-    if (!node) return;
-    node->cmd = malloc(len);
-    if (!node->cmd) { free(node); return; }
-    memcpy(node->cmd, cmd, len);
-    node->len = len;
-    node->next = NULL;
-    if (q->tail) q->tail->next = node;
-    else q->head = node;
-    q->tail = node;
-    q->count++;
-}
+static replica_state_e state = OFFLINE;
+static int sockfd = -1;

-static void queue_send_and_clear(struct pending_queue *q, int sock) {
-    struct cmd_node *node = q->head;
-    int sent_count = 0;
-    while (node) {
-        if (send(sock, node->cmd, node->len, 0) > 0) {
-            sent_count++;
-        }
-        struct cmd_node *tmp = node;
-        node = node->next;
-        free(tmp->cmd);
-        free(tmp);
-    }
-    if (sent_count > 0) {
-        printf("[QUEUE] Sent %d commands to slave\n", sent_count);
-    }
-    queue_init(q);
-}
+static char peer_ip[MAX_IP_LEN];
+static int peer_port;
+static __u64 peer_seq;

-static void queue_free(struct pending_queue *q) {
-    struct cmd_node *node = q->head;
-    while (node) {
-        struct cmd_node *tmp = node;
-        node = node->next;
-        free(tmp->cmd);
-        free(tmp);
-    }
-    queue_init(q);
-}
+static struct pending_queue pending = {
+    .head = NULL,
+    .tail = NULL,
+    .count = 0,
+};

-static int send_file(int sock, const char *path) {
-    FILE *fp = fopen(path, "rb");
-    if (!fp) {
-        printf("[ERROR] Failed to open file: %s\n", path);
-        return -1;
-    }
-    
-    char buf[4096];
-    size_t n, total = 0;
-    while ((n = fread(buf, 1, sizeof(buf), fp)) > 0) {
-        if (send(sock, buf, n, 0) < 0) { 
-            fclose(fp); 
-            printf("[ERROR] Failed to send file: %s (sent %zu bytes)\n", path, total);
-            return -1; 
-        }
-        total += n;
-    }
-    fclose(fp);
-    printf("[FILE] Sent %s (%zu bytes)\n", path, total);
-    return 0;
-}
-
-// 全局状态（单 Slave 简化）
-static enum state current_state = NOSLAVE;
-static char slave_ip[16] = {0};
-static int slave_port = 0;
-static char array_file[128] = {0};
-static char rbtree_file[128] = {0};
-static char hash_file[128] = {0};
-static struct pending_queue pending;
-static int slave_sock = -1;  // 连接 Slave 的 socket
-
-// 连接 Slave
-static int connect_slave() {
-    int sock = socket(AF_INET, SOCK_STREAM, 0);
-    if (sock < 0) return -1;
-    struct sockaddr_in addr;
-    memset(&addr, 0, sizeof(addr));
-    addr.sin_family = AF_INET;
-    addr.sin_port = htons(slave_port);
-    inet_pton(AF_INET, slave_ip, &addr.sin_addr);
-    if (connect(sock, (struct sockaddr *)&addr, sizeof(addr)) < 0) {
-        close(sock);
-        return -1;
-    }
-    return sock;
-}
-
-static void handle_event(void *ctx, int cpu, void *data, __u32 size) {
-    struct event *ev = (struct event *)data;
-
-    switch (ev->type) {
-        case EVENT_CREATE_SNAPSHOT_ASYNC:
-			printf("[EVENT] Type: CREATE_SNAPSHOT_ASYNC\n");
-            printf("[EVENT] Slave IP: %s, Port: %u\n", ev->data.sync.ip, ev->data.sync.port);
-
-            if (current_state == NOSLAVE) {
-                current_state = START;
-                strncpy(slave_ip, ev->data.sync.ip, sizeof(slave_ip));
-                slave_port = ev->data.sync.port;
-                queue_init(&pending);
-                slave_sock = connect_slave();  // 连接 Slave
-                if (slave_sock < 0) {
-                    printf("Failed to connect to Slave %s:%d\n", slave_ip, slave_port);
-                    current_state = NOSLAVE;
-                }
-            }
-            break;
-        case EVENT_COMPLETED_CMD:
-			printf("[EVENT] Type: COMPLETED_CMD\n");
-            printf("[EVENT] Command length: %llu bytes\n", ev->data.cmd.len);
-
-            if (current_state != NOSLAVE) {
-                queue_push(&pending, ev->data.cmd.cmd, ev->data.cmd.len);
-            }
-            break;
-        case EVENT_CREATE_SNAPSHOT_OK:
-			printf("[EVENT] Type: CREATE_SNAPSHOT_OK\n");
-            printf("[EVENT] Array file: %s\n", ev->data.ok.array_file);
-            printf("[EVENT] RBTree file: %s\n", ev->data.ok.rbtree_file);
-            printf("[EVENT] Hash file: %s\n", ev->data.ok.hash_file);
-
-            if (current_state == START) {
-                current_state = DONE;
-                strncpy(array_file, ev->data.ok.array_file, sizeof(array_file));
-                strncpy(rbtree_file, ev->data.ok.rbtree_file, sizeof(rbtree_file));
-                strncpy(hash_file, ev->data.ok.hash_file, sizeof(hash_file));
-            }
-            break;
-    }
-}
-
-
-static void lost_event(void *ctx, int cpu, __u64 cnt) {
-    printf("Lost %llu events\n", cnt);
-}
-
-static int libbpf_print_fn(enum libbpf_print_level level, const char *format, va_list args)
+/* ================= pending 队列操作 ================= */
+static void pending_free()
 {
-	return vfprintf(stderr, format, args);
+    struct pending_queue *q = &pending;
+    struct cmd_node *cur = q->head;
+    while (cur) {
+        struct cmd_node *tmp = cur;
+        cur = cur->next;
+        free(tmp->cmd);
+        free(tmp);
+    }
+    q->head = q->tail = NULL;
+    q->count = 0;
 }

+static void pending_push(__u64 seq, __u32 len, const uint8_t *cmd)
+{
+    struct cmd_node *node = malloc(sizeof(*node));
+    if (!node)
+        return;
+
+    node->cmd = malloc(len);
+    if (!node->cmd) {
+        free(node);
+        return;
+    }
+
+    memcpy(node->cmd, cmd, len);
+    node->seq  = seq;
+    node->len  = len;
+    node->next = NULL;
+
+    if (!pending.tail) {
+        pending.head = pending.tail = node;
+    } else {
+        pending.tail->next = node;
+        pending.tail = node;
+    }
+
+    pending.count++;
+}
+
+static void pending_gc(__u64 min_seq)
+{
+    struct cmd_node *cur = pending.head;
+
+    int n = pending.count;
+    while (cur && cur->seq < min_seq) {
+        struct cmd_node *tmp = cur;
+        cur = cur->next;
+
+        free(tmp->cmd);
+        free(tmp);
+        pending.count--;
+    }
+
+    printf("gc:%d\n", n-pending.count);
+
+    pending.head = cur;
+    if (!cur)
+        pending.tail = NULL;
+}
+
+static void pending_send_one(struct cmd_node *node){
+    int rt = send(sockfd, node->cmd, node->len, 0);
+    printf("send seq:%lld, rt=%d\n", node->seq, rt);
+}
+
+static void pending_send_all(void)
+{
+    struct cmd_node *cur = pending.head;
+
+    while (cur) {
+        pending_send_one(cur);
+        cur = cur->next;
+    }
+}
+
+/* ================= 网络逻辑 ================= */
+static void try_connect(void)
+{
+    if(sockfd > 0){
+        close(sockfd);
+        sockfd = -1;
+    }
+
+    struct sockaddr_in addr = {};
+    int i = 0;
+
+    addr.sin_family = AF_INET;
+    addr.sin_port = htons(peer_port);
+    inet_pton(AF_INET, peer_ip, &addr.sin_addr);
+
+    for(i = 0;i < 10; ++ i){
+        sockfd = socket(AF_INET, SOCK_STREAM, 0);
+        if (sockfd < 0) {
+            perror("socket");
+            return;
+        }
+
+        printf("connect try %d...\n", i + 1);
+        if (connect(sockfd, (struct sockaddr *)&addr, sizeof(addr)) == 0) {
+            printf("connect success: %s:%d\n", peer_ip, peer_port);
+            state = ONLINE;
+            pending_send_all();
+            return;
+        }
+
+        perror("connect");
+        close(sockfd);
+        sockfd = -1;
+
+        sleep(1);
+    }
+
+    printf("connect failed after 10 retries\n");
+}
+
+
+/* ================= perf buffer 回调 ================= */
+static void handle_event(void *ctx, int cpu, void *data, __u32 size)
+{
+    struct replica_event *evt = data;
+
+    switch (evt->type) {
+
+    case EVENT_SSYNC:
+        strncpy(peer_ip, evt->sync.ip, sizeof(peer_ip));
+        peer_port = evt->sync.port;
+        peer_seq  = evt->sync.seq;
+        printf("SSYNC [seq:%lld], [%s:%d]\n", peer_seq, peer_ip, peer_port);
+
+        state = OFFLINE;
+        pending_gc(peer_seq);
+        break;
+
+    case EVENT_COMPLETED_CMD:
+        // printf("CMD [seq:%lld], cmd:\n[\n%s]\n", evt->complete.seq, evt->complete.cmd);
+        pending_push(evt->complete.seq,
+                     evt->complete.len,
+                     evt->complete.cmd);
+
+        if (state == ONLINE && pending.tail) {
+            struct cmd_node *n = pending.tail;
+            pending_send_one(n);
+        }
+        break;
+
+    case EVENT_SREADY:
+        printf("SREADY \n");
+        if (state == OFFLINE)
+            try_connect();
+        break;
+    }
+}
+
+/* ================= main ================= */
 int main(int argc, char **argv)
 {
 	struct replica_bpf *skel;
+    struct perf_buffer *pb = NULL;
 	int err;

-	/* Set up libbpf errors and debug info callback */
-	libbpf_set_print(libbpf_print_fn);
-
 	/* Open BPF application */
 	skel = replica_bpf__open();
 	if (!skel) {
@@ -190,9 +217,6 @@ int main(int argc, char **argv)
 		return 1;
 	}

-	/* ensure BPF program only handles write() syscalls from our process */
-	skel->bss->my_pid = getpid();
-
 	/* Load & verify BPF programs */
 	err = replica_bpf__load(skel);
 	if (err) {
@@ -211,47 +235,22 @@ int main(int argc, char **argv)
 	       "to see output of the BPF programs.\n");


-	struct perf_buffer *pb = perf_buffer__new(bpf_map__fd(skel->maps.channel), 8, handle_event, lost_event, NULL, NULL);
+	pb = perf_buffer__new(bpf_map__fd(skel->maps.events), 8,
+                 handle_event, NULL, NULL, NULL);
+
 	if(!pb){
 		goto cleanup;
 	}

-#if 0
-	while(1){
-		perf_buffer__poll(pb, 1000);
-	}
-#else 
 	while (1) {
        perf_buffer__poll(pb, 1000);  // 处理事件
-
-        // 循环中检查状态并发送
-        if (current_state == DONE && slave_sock >= 0) {
-            // 发送快照文件
-            if (send_file(slave_sock, array_file) == 0 &&
-                send_file(slave_sock, rbtree_file) == 0 &&
-                send_file(slave_sock, hash_file) == 0) {
-                current_state = ONLINE;
-                printf("Snapshot sent, state to ONLINE\n");
-            } else {
-                printf("Failed to send snapshot\n");
-                current_state = NOSLAVE;
-                close(slave_sock);
-                slave_sock = -1;
-            }
-        }
-
-        if (current_state == ONLINE && slave_sock >= 0) {
-            // 发送 pending
-            queue_send_and_clear(&pending, slave_sock);
-        }
    }
-#endif

 	perf_buffer__free(pb);

 cleanup:
-	queue_free(&pending);
-	if (slave_sock >= 0) close(slave_sock);
+	pending_free();
+	if (sockfd >= 0) close(sockfd);
 	replica_bpf__destroy(skel);
 	return -err;
 }
--- a/ebpf/c/replica.h
+++ b/ebpf/c/replica.h
@@ -1,36 +1,37 @@
 #ifndef __REPLICA_H__
 #define __REPLICA_H__

+
+#define MAX_CMD_LEN 256
+#define MAX_IP_LEN  64
+
 enum event_type {
-    EVENT_CREATE_SNAPSHOT_ASYNC,
-    EVENT_CREATE_SNAPSHOT_OK,
-    EVENT_CREATE_SNAPSHOT_READY,
-    EVENT_COMPLETED_CMD
+    EVENT_COMPLETED_CMD,
+    EVENT_SSYNC,
+    EVENT_SREADY,
 };

-struct event {
-    enum event_type type;
+struct complete_cmd_evt {
+    __u64 seq;
+    __u32 len;
+    __u8  cmd[MAX_CMD_LEN];
+};
+
+struct sync_evt {
+    __u64 seq;
+    char  ip[MAX_IP_LEN];
+    __s32 port;
+};
+
+struct replica_event {
+    __u32 type;
+    __u32 _pad;
+
    union {
-        struct {
-            char ip[16];
-            __u32 port;
-        } sync;
-        struct {
-            __u8 cmd[256];
-            __u64 len;
-        } cmd;
-        struct {
-            char array_file[128];
-            char rbtree_file[128];
-            char hash_file[128];
-        } ok;
-    } data;
+        struct complete_cmd_evt complete;
+        struct sync_evt sync;
+    };
 };

-enum state {
-    NOSLAVE,
-    PREPARING,
-    ONLINE
-};

 #endif
--- a/kvs_array_bin.c
+++ b/kvs_array_bin.c
@@ -237,13 +237,13 @@ int kvs_array_save(iouring_ctx_t *uring, kvs_array_t *inst, const char* filename
        task_t *t = submit_write(uring, fd, bufs, lens, count, current_off);

 		if (!t) { close(fd); return -4; }
-        int res = task_wait(t);
-        task_destroy(t);
+        // int res = task_wait(t);
+        // task_destroy(t);

-        if (res < 0) { 
-            close(fd); 
-            return -5; 
-        }
+        // if (res < 0) { 
+        //     close(fd); 
+        //     return -5; 
+        // }
        
        current_off += (off_t) total;
 	}
--- a/kvs_hash_bin.c
+++ b/kvs_hash_bin.c
@@ -310,13 +310,13 @@ int kvs_hash_save(iouring_ctx_t *uring, kvs_hash_t *inst, const char* filename){
 			task_t *t = submit_write(uring, fd, bufs, lens, count, current_off);

 			if (!t) { close(fd); return -4; }
-			int res = task_wait(t);
-			task_destroy(t);
+			// int res = task_wait(t);
+			// task_destroy(t);

-			if (res < 0) { 
-				close(fd); 
-				return -5; 
-			}
+			// if (res < 0) { 
+			// 	close(fd); 
+			// 	return -5; 
+			// }
 			
 			current_off += (off_t) total;
 		}
--- a/kvs_protocol_resp.c
+++ b/kvs_protocol_resp.c
@@ -14,6 +14,8 @@ extern kvs_rbtree_t global_rbtree;
 extern kvs_hash_t global_hash;
 #endif

+extern unsigned long long global_seq;
+
 static int need(const uint8_t *p, const uint8_t *end, size_t n) {
    return (p + n <= end) ? 0 : -1;
 }
@@ -543,10 +545,12 @@ int resp_dispatch(const resp_cmd_t *cmd, resp_value_t *out_value) {
        return 0;
    }
 	case KVS_CMD_SSYNC:
+        __ssync(cmd->argv[1].ptr, cmd->argv[1].len, atoi(cmd->argv[2].ptr), global_seq);
        kvs_create_snapshot_async(cmd->argv[1].ptr, atoi(cmd->argv[2].ptr));
 		*out_value = resp_simple("OK");
 		return 0;
    case KVS_CMD_SREADY:
+        __sready();
 		*out_value = resp_simple("OK");
 		return 0;
 	default:
@@ -556,3 +560,18 @@ int resp_dispatch(const resp_cmd_t *cmd, resp_value_t *out_value) {
    *out_value = resp_error("ERR unknown command");
    return 0;
 }
+
+void __ssync(const uint8_t *ip, uint32_t ip_len, int port, unsigned long long seq){
+    // hook 在这里，必须等待 ebpf实时同步进程 状态切换为 PREPARING 才允许返回
+    // 不这样做的话：快照立刻被创建，执行下一条命令(CmdX)，如果此刻 ebpf 还没有被置为 PREPARING，即没有开始记录需要转发的命令
+    // 会导致：快照里没有(CmdX)，ebpf实时同步进程里也没有(CmdX)
+    // 要怎么做？
+    // master 在内部维护seq，ebpf 从启动就一直记录数据。ebpf探测 __ssync 的seq来确定从哪里开始发送。
+    // ebpf: __ssync            SSYNC ip port seq,  知道往哪里发，从哪里开始发
+    // ebpf: __sready           SREADY,             可以开始发了，发往(ip,port)
+    // ebpf: __complete_cmd     cmd,len,seq,        记录命令
+}
+
+void __sready(){
+
+}
--- a/kvs_protocol_resp.h
+++ b/kvs_protocol_resp.h
@@ -90,4 +90,7 @@ resp_value_t resp_nil(void);             /* $-1\r\n */
 */
 int resp_dispatch(const resp_cmd_t *cmd, resp_value_t *out_value);

+void __ssync(const uint8_t *ip, uint32_t ip_len, int port, unsigned long long seq);
+void __sready();
+
 #endif
--- a/kvs_rbtree_bin.c
+++ b/kvs_rbtree_bin.c
@@ -507,12 +507,12 @@ static int kvs_rbtree_save_node(iouring_ctx_t *uring, int fd, off_t *current_off

 	
 	if (!t) { return -4; }
-	int res = task_wait(t);
+	// int res = task_wait(t);
 	task_destroy(t);

-	if (res < 0) { 
-		return -5; 
-	}
+	// if (res < 0) { 
+	// 	return -5; 
+	// }
 	
 	*current_off += (off_t) total;

--- a/kvs_slave.c
+++ b/kvs_slave.c
@@ -192,17 +192,17 @@ int slave_bootstrap(
    printf("Sent SREADY to master\n");

    // 6. 接收回包+OK\r\n
-    if (recv_exact(master_fd, resp, 5) < 0) {
-        fprintf(stderr, "Failed to receive final OK from master\n");
-        goto cleanup;
-    }
+    // if (recv_exact(master_fd, resp, 5) < 0) {
+    //     fprintf(stderr, "Failed to receive final OK from master\n");
+    //     goto cleanup;
+    // }
    
-    if (memcmp(resp, "+OK\r\n", 5) != 0) {
-        fprintf(stderr, "Unexpected final response from master: %.5s\n", resp);
-        goto cleanup;
-    }
+    // if (memcmp(resp, "+OK\r\n", 5) != 0) {
+    //     fprintf(stderr, "Unexpected final response from master: %.5s\n", resp);
+    //     goto cleanup;
+    // }

-    printf("Received final OK from master, bootstrap complete\n");
+    // printf("Received final OK from master, bootstrap complete\n");
    
    ret = 0;

--- a/kvstore.c
+++ b/kvstore.c
@@ -21,16 +21,15 @@

 extern int slave_bootstrap(const char *listen_ip, int listen_port, const char *master_ip, int master_port);

-#if MEMORY_SELECT_MALLOC == MEMORY_USE_MYMALLOC
 extern mp_pool_t global_mempool;
-#endif

 AppConfig global_cfg;
 iouring_ctx_t global_uring_ctx;
+unsigned long long global_seq;

 extern int global_oplog_fd;

-void __completed_cmd(const uint8_t *cmd, size_t len){
+void __completed_cmd(const uint8_t *cmd, size_t len, unsigned long long seq){

 }

@@ -69,7 +68,8 @@ int kvs_protocol(struct conn* conn){

 		int dr = resp_dispatch(&cmd, &val);

-		__completed_cmd(p, len);
+		__completed_cmd(p, len, global_seq);
+		global_seq ++;


        /*
@@ -186,9 +186,10 @@ void dest_kvengine(void) {
 }

 void init_memory_pool(AppConfig *cfg){
-#if MEMORY_SELECT_MALLOC == MEMORY_USE_MYMALLOC
-	mp_create(&global_mempool);
-#endif
+
+	if(cfg->allocator == ALLOC_MYPOOL)
+		mp_create(&global_mempool);
+		
 	kvs_set_memleak_detect(cfg->leak_mode);
 	kvs_set_alloc_type(cfg->allocator);

@@ -196,7 +197,7 @@ void init_memory_pool(AppConfig *cfg){
 }

 void dest_memory_pool(void){
-#if MEMORY_SELECT_MALLOC == MEMORY_USE_MYMALLOC
+#if MEMORY_USE_MYMALLOC
 	mp_destroy(&global_mempool);
 #endif 
 }
@@ -278,9 +279,10 @@ int main(int argc, char *argv[]) {
 		printf("Init Config error");
 		return -1;
 	} 
+	global_seq = 0;

+	init_memory_pool(&global_cfg);
 	init_data_file(&global_cfg);
-
 	init_disk_uring(&global_uring_ctx);
 	
 	int port = global_cfg.port;
@@ -296,9 +298,6 @@ int main(int argc, char *argv[]) {
 		
 	}

-
-
-	init_memory_pool(&global_cfg);
 	init_kvengine();
 	
 #if (NETWORK_SELECT == NETWORK_REACTOR)
--- a/kvstore.h
+++ b/kvstore.h
@@ -279,7 +279,8 @@ extern kvs_rbtree_t global_rbtree;
 extern kvs_hash_t global_hash;
 #endif

-void __completed_cmd(const uint8_t *cmd, size_t len);
+void __completed_cmd(const uint8_t *cmd, size_t len, unsigned long long seq);
+

 #endif

--- a/memory/alloc_dispatch.c
+++ b/memory/alloc_dispatch.c
@@ -9,7 +9,7 @@

 mp_pool_t global_mempool;

-static atomic_int g_memory_mode         = ATOMIC_VAR_INIT(MEMLEAK_DETECT_OFF);
+static atomic_int g_memory_mode         = ATOMIC_VAR_INIT(ALLOC_OTHER);
 static atomic_int g_memleak_detect_mode = ATOMIC_VAR_INIT(MEMLEAK_DETECT_OFF);

 // 设置内存池类型
@@ -36,7 +36,7 @@ MemLeakDetectMode kvs_get_memleak_detect(void) {

 void *kvs_malloc_impl(size_t size){

-    switch (atomic_load(&g_memleak_detect_mode)){
+    switch (atomic_load(&g_memory_mode)){
        case ALLOC_MALLOC:
            return malloc(size);
        case ALLOC_MYPOOL:
@@ -51,7 +51,7 @@ void *kvs_malloc_impl(size_t size){
 }

 void kvs_free_impl(void *ptr) {
-    switch (atomic_load(&g_memleak_detect_mode)){
+    switch (atomic_load(&g_memory_mode)){
        case ALLOC_MALLOC:
 	        free(ptr);
            break;
@@ -75,20 +75,20 @@ void kvs_free_impl(void *ptr) {
 void *nMalloc(size_t size, const char * filename, const char *func, int line){
    void *ptr = kvs_malloc_impl(size);

-    if(atomic_load(&g_memleak_detect_mode) == MEMLEAK_DETECT_ON) {
-        char buff[128];
-        snprintf(buff, 128, "./mem_leak/%p.mem", ptr);
-        FILE* fp = fopen(buff, "w");
-        if(!fp){
-            kvs_free(ptr);
-            return NULL;
-        }
+    // if(atomic_load(&g_memleak_detect_mode) == MEMLEAK_DETECT_ON) {
+    //     char buff[128];
+    //     snprintf(buff, 128, "./mem_leak/%p.mem", ptr);
+    //     FILE* fp = fopen(buff, "w");
+    //     if(!fp){
+    //         kvs_free(ptr);
+    //         return NULL;
+    //     }

-        fprintf(fp, "[+] [%s:%d:%s] [%p:%ld]\n", filename, line, func, ptr, size);
-        fflush(fp);
-        fclose(fp);
+    //     fprintf(fp, "[+] [%s:%d:%s] [%p:%ld]\n", filename, line, func, ptr, size);
+    //     fflush(fp);
+    //     fclose(fp);

-    }
+    // }
    return ptr;
 }

@@ -97,15 +97,15 @@ void nFree(void *ptr, const char * filename, const char *func, int line){
        return ;
    }
    
-    if(atomic_load(&g_memleak_detect_mode) == MEMLEAK_DETECT_ON) {    
-        char buff[128];
-        snprintf(buff, 128, "./mem_leak/%p.mem", ptr);
+    // if(atomic_load(&g_memleak_detect_mode) == MEMLEAK_DETECT_ON) {    
+    //     char buff[128];
+    //     snprintf(buff, 128, "./mem_leak/%p.mem", ptr);

-        if(unlink(buff) < 0) {
-            return ;
-        }
+    //     if(unlink(buff) < 0) {
+    //         return ;
+    //     }

-    }
+    // }

    kvs_free_impl(ptr);
 }
--- a/memory/mempool.c
+++ b/memory/mempool.c
@@ -74,7 +74,7 @@ static mp_page_t* mp_page_create(mp_bucket_t *owner){
    pg->prev = NULL;
    pg->next = NULL;

-    bitmap_clear_all(pg->bitmap, 8);
+    bitmap_clear_all(pg->bitmap, 20);

    char *p = (char*)page_payload(pg);
    for(uint16_t i = 0;i < cap - 1; ++ i){
--- a/memory/mempool.h
+++ b/memory/mempool.h
@@ -6,11 +6,12 @@
 #include <stdlib.h>
 #include <string.h>

-#define MEMPOOL_PAGE_SIZE           4096
+// #define MEMPOOL_PAGE_SIZE           4096
+#define MEMPOOL_PAGE_SIZE           (1024*8)
 #define MEMPOOL_BLOCK_MAX_SIZE      512
 #define MEMPOOL_ALIGNMENT           8
 #define MEMPOOL_NUM_CLASSES         (MEMPOOL_BLOCK_MAX_SIZE / MEMPOOL_ALIGNMENT)
-#define MEMPOOL_CACHE_PAGE          2
+#define MEMPOOL_CACHE_PAGE          4

 typedef struct mp_page_s   mp_page_t;
 typedef struct mp_bucket_s mp_bucket_t;
@@ -34,7 +35,7 @@ struct mp_page_s{
    uint16_t        free_count;
    uint16_t        capacity;

-    uint64_t        bitmap[8];  // 最多支持 512 个块 (64*8)
+    uint64_t        bitmap[20];  // 最多支持 512/1280 个块 (64*20)
 };

 struct mp_bucket_s{
--- a/reactor.c
+++ b/reactor.c
@@ -336,6 +336,12 @@ int r_init_server(unsigned short port) {

 	int sockfd = socket(AF_INET, SOCK_STREAM, 0);

+	int opt = 1;
+    if (setsockopt(sockfd, SOL_SOCKET, SO_REUSEADDR, &opt, sizeof(opt)) < 0) {
+        perror("setsockopt");
+		close(sockfd);
+    }
+
 	struct sockaddr_in servaddr;
 	servaddr.sin_family = AF_INET;
 	servaddr.sin_addr.s_addr = htonl(INADDR_ANY); // 0.0.0.0
--- a/test-redis/test.c
+++ b/test-redis/test.c
@@ -5,7 +5,8 @@
 #include <hiredis/hiredis.h>

 #define TIME_SUB_MS(tv1, tv2)  ((tv1.tv_sec - tv2.tv_sec) * 1000 + (tv1.tv_usec - tv2.tv_usec) / 1000)
-
+// #define PRINT   printf
+#define PRINT   

 static void die(redisContext *c, const char *msg) {
    fprintf(stderr, "%s: %s\n", msg, c && c->err ? c->errstr : "unknown");
@@ -38,6 +39,7 @@ static void must_bulk_eq(redisReply *r, const void *buf, size_t n, const char *w
    if (!r) { fprintf(stderr, "%s: reply null\n", what); exit(1); }
    if (r->type != REDIS_REPLY_STRING || r->len != n || memcmp(r->str, buf, n) != 0) {
        fprintf(stderr, "%s: bulk mismatch. type=%d len=%zu\n", what, r->type, r->len);
+        fprintf(stderr, "expect:%s, truely:%s\n", (const char*)buf, r->str);
        freeReplyObject(r);
        exit(1);
    }
@@ -111,17 +113,18 @@ void pipline_set_test(redisContext *c, int start, int countN, const char *op){
            die(c, "redisAppendCommand SET failed");            
        }

-        if(i%10000 == 0) printf("%d\n", i);
-
+        if(i%10000 == 0) PRINT("SEND: %d\n", i);
    }
    /* 再一次性把 N 个回复读出来 */
    for (int i = start; i < end; i++) {
        redisReply *r = NULL;
        if (redisGetReply(c, (void**)&r) != REDIS_OK || !r) die(c, "redisGetReply SET failed");
        must_ok(r, "pipeline SET reply");
+
+        if(i%10000 == 0) PRINT("RECV: %d\n", i);
    }

-    printf("[OK] SET pipeline batch %d\n", N);
+    PRINT("[OK] SET pipeline batch %d\n", N);
 }

 void pipline_get_test(redisContext *c, int start, int countN, const char *op){
@@ -138,7 +141,7 @@ void pipline_get_test(redisContext *c, int start, int countN, const char *op){
            die(c, "redisAppendCommand GET failed");
        }
        
-        if(i%10000 == 0) printf("%d\n", i);
+        if(i%10000 == 0) PRINT("SEND: %d\n", i);
    }
    for (int i = start; i < end; i++) {
        redisReply *r = NULL;
@@ -146,9 +149,11 @@ void pipline_get_test(redisContext *c, int start, int countN, const char *op){
        char expect[64];
        int en = snprintf(expect, sizeof(expect), "v:%d", i);
        must_bulk_eq(r, expect, (size_t)en, "pipeline GET reply");
+        
+        if(i%10000 == 0) PRINT("RECV: %d\n", i);
    }

-    printf("[OK] GET pipeline batch %d\n", N);
+    PRINT("[OK] GET pipeline batch %d\n", N);
 }

 void pipline_del_test(redisContext *c, int start, int countN, const char *op){
@@ -165,17 +170,44 @@ void pipline_del_test(redisContext *c, int start, int countN, const char *op){
            die(c, "redisAppendCommand DEL failed");
        }

-        if(i%10000 == 0) printf("%d\n", i);
+        if(i%10000 == 0) PRINT("SEND: %d\n", i);
    }
    for (int i = start; i < end; i++) {
        redisReply *r = NULL;
        if (redisGetReply(c, (void**)&r) != REDIS_OK || !r) die(c, "redisGetReply DEL failed");
        freeReplyObject(r); /* DEL 返回 int，这里不强制检查 */
+
+        if(i%10000 == 0) PRINT("RECV: %d\n", i);
    }

-    printf("[OK] DEL pipeline batch %d\n", N);
+    PRINT("[OK] DEL pipeline batch %d\n", N);
 }

+long long test_nopersist_noreplica(redisContext *c, int rounds, long long batch_size){
+    struct timeval tv_begin, tv_end;
+    gettimeofday(&tv_begin, NULL);
+
+    long long total_ops  = batch_size*rounds;
+
+    for(int i = 0;i < total_ops ; i += batch_size){
+        pipline_set_test(c, i, batch_size, "RSET");
+    }
+
+    for(int i = 0;i < total_ops ; i += batch_size){
+        pipline_get_test(c, i, batch_size, "RGET");
+    }
+
+    for(int i = 0;i < total_ops ; i += batch_size){
+        pipline_del_test(c, i, batch_size, "RDEL");
+    }
+
+    gettimeofday(&tv_end, NULL);
+    int time_used = TIME_SUB_MS(tv_end, tv_begin);
+    long long qps = total_ops *3*1000/time_used;
+    printf("BATCH (N=%lld) --> time_used=%d ms, qps=%lld\n", total_ops *3, time_used, qps);
+
+    return qps;
+}

 int main(int argc, char **argv) {
    if(argc < 4) {
@@ -197,29 +229,18 @@ int main(int argc, char **argv) {
    }else if(mode == 1){
        basic_command_test(c);
    }else if(mode == 3){
-        struct timeval tv_begin, tv_end;
-        gettimeofday(&tv_begin, NULL);
-        long long ops = 100000;
+        int rounds = 10;
+        long long batch_size = 100000;
+        int testrounds = 50;
+        long long total_qps = 0;

-        pipline_set_test(c, 0, ops, "RSET");
-
-        gettimeofday(&tv_end, NULL);
-        int time_used = TIME_SUB_MS(tv_end, tv_begin);
-        long long qps = ops*1000/time_used;
-        printf("BATCH (N=%lld) --> time_used=%d ms, qps=%lld\n",
-           ops, time_used, qps);
+        for(int i = 0;i < testrounds; ++ i){
+            total_qps += test_nopersist_noreplica(c, rounds, batch_size);
+        }
+        printf("average qps:%lld\n", total_qps/testrounds);
    }else if(mode == 4){
-        struct timeval tv_begin, tv_end;
-        gettimeofday(&tv_begin, NULL);      
-        long long ops = 100000;

-        pipline_del_test(c, 0, ops, "RDEL");
-
-        gettimeofday(&tv_end, NULL);
-        int time_used = TIME_SUB_MS(tv_end, tv_begin);
-        long long qps = ops*1000/time_used;
-        printf("BATCH (N=%lld) --> time_used=%d ms, qps=%lld\n",
-           ops, time_used, qps);
+    }else if(mode == 5){

    }else if(mode == 10){
        pipline_set_test(c, 0, 1000, "SET");