diff --git a/src/inc/taosdef.h b/src/inc/taosdef.h
index fa8af2c67e5dff0cbc0a66a4fca3bfa9f84cbe26..875092b88da1aa03b7eb017e14fff3148f3644a8 100644
--- a/src/inc/taosdef.h
+++ b/src/inc/taosdef.h
@@ -431,6 +431,8 @@ void tsDataSwap(void *pLeft, void *pRight, int32_t type, int32_t size, void* buf
 #define TSDB_PORT_HTTP       11 
 #define TSDB_PORT_ARBITRATOR 12
 
+#define TSDB_MAX_WAL_SIZE    (1024*1024)
+
 typedef enum {
   TAOS_QTYPE_RPC   = 0,
   TAOS_QTYPE_FWD   = 1,
diff --git a/src/inc/taoserror.h b/src/inc/taoserror.h
index 01891234efeae1972a3884c57b99d204a335e7d5..1919747a0b019239e5d6b717921c6a957728d455 100644
--- a/src/inc/taoserror.h
+++ b/src/inc/taoserror.h
@@ -237,7 +237,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_QRY_NOT_READY,                0, 0x0707, "Query not
 TAOS_DEFINE_ERROR(TSDB_CODE_QRY_HAS_RSP,                  0, 0x0708, "Query should response")
 TAOS_DEFINE_ERROR(TSDB_CODE_QRY_IN_EXEC,                  0, 0x0709, "Multiple retrieval of this query")
 TAOS_DEFINE_ERROR(TSDB_CODE_QRY_TOO_MANY_TIMEWINDOW,      0, 0x070A, "Too many time window in query")
-TAOS_DEFINE_ERROR(TSDB_CODE_QRY_NOT_ENOUGH_BUFFER,      0, 0x070B, "Query buffer limit has reached")
+TAOS_DEFINE_ERROR(TSDB_CODE_QRY_NOT_ENOUGH_BUFFER,        0, 0x070B, "Query buffer limit has reached")
 
 // grant
 TAOS_DEFINE_ERROR(TSDB_CODE_GRANT_EXPIRED,                0, 0x0800, "License expired")
@@ -261,6 +261,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_SYN_INVALID_VERSION,          0, 0x0902, "Invalid Sy
 // wal
 TAOS_DEFINE_ERROR(TSDB_CODE_WAL_APP_ERROR,                0, 0x1000, "Unexpected generic error in wal")
 TAOS_DEFINE_ERROR(TSDB_CODE_WAL_FILE_CORRUPTED,           0, 0x1001, "WAL file is corrupted")
+TAOS_DEFINE_ERROR(TSDB_CODE_WAL_SIZE_LIMIT,               0, 0x1002, "WAL size exceeds limit")
 
 // http
 TAOS_DEFINE_ERROR(TSDB_CODE_HTTP_SERVER_OFFLINE,          0, 0x1100, "http server is not onlin")
diff --git a/src/inc/twal.h b/src/inc/twal.h
index c32bb870213fbcb183af599b0ac94b37f54e8906..b85377d8d43521f4927b958d1839a9e565431c6a 100644
--- a/src/inc/twal.h
+++ b/src/inc/twal.h
@@ -59,6 +59,7 @@ int32_t walAlter(twalh pWal, SWalCfg *pCfg);
 void    walStop(twalh);
 void    walClose(twalh);
 int32_t walRenew(twalh);
+void    walRemoveOldFiles(twalh);
 int32_t walWrite(twalh, SWalHead *);
 void    walFsync(twalh, bool forceFsync);
 int32_t walRestore(twalh, void *pVnode, FWalWrite writeFp);
diff --git a/src/sync/inc/syncInt.h b/src/sync/inc/syncInt.h
index 240b401bdaef252e36109490658b39ede749a8fe..93c6efc20a8d79a7d6436680f01cee87dadda853 100644
--- a/src/sync/inc/syncInt.h
+++ b/src/sync/inc/syncInt.h
@@ -35,6 +35,8 @@ extern "C" {
 #define TAOS_SMSG_SYNC_MUST    6
 #define TAOS_SMSG_STATUS       7
 
+#define SYNC_MAX_SIZE (TSDB_MAX_WAL_SIZE + sizeof(SWalHead) + sizeof(SSyncHead) + 16)
+
 #define nodeRole    pNode->peerInfo[pNode->selfIndex]->role
 #define nodeVersion pNode->peerInfo[pNode->selfIndex]->version
 #define nodeSStatus pNode->peerInfo[pNode->selfIndex]->sstatus
diff --git a/src/sync/src/syncMain.c b/src/sync/src/syncMain.c
index 9dcd0fd632396baf5f8bf68c3df24dae72d45a02..ef1ada4c2efd4967bdfee871596a61806db46a15 100644
--- a/src/sync/src/syncMain.c
+++ b/src/sync/src/syncMain.c
@@ -79,7 +79,7 @@ int32_t syncInit() {
   info.numOfThreads = tsSyncTcpThreads;
   info.serverIp = 0;
   info.port = tsSyncPort;
-  info.bufferSize = 640000;
+  info.bufferSize = SYNC_MAX_SIZE;
   info.processBrokenLink = syncProcessBrokenLink;
   info.processIncomingMsg = syncProcessPeerMsg;
   info.processIncomingConn = syncProcessIncommingConnection;
@@ -850,7 +850,7 @@ static void syncProcessForwardFromPeer(char *cont, SSyncPeer *pPeer) {
   SSyncNode *pNode = pPeer->pSyncNode;
   SWalHead * pHead = (SWalHead *)cont;
 
-  sDebug("%s, forward is received, ver:%" PRIu64, pPeer->id, pHead->version);
+  sDebug("%s, forward is received, hver:%" PRIu64 ", len:%d", pPeer->id, pHead->version, pHead->len);
 
   if (nodeRole == TAOS_SYNC_ROLE_SLAVE) {
     // nodeVersion = pHead->version;
@@ -859,7 +859,7 @@ static void syncProcessForwardFromPeer(char *cont, SSyncPeer *pPeer) {
     if (nodeSStatus != TAOS_SYNC_STATUS_INIT) {
       syncSaveIntoBuffer(pPeer, pHead);
     } else {
-      sError("%s, forward discarded, ver:%" PRIu64, pPeer->id, pHead->version);
+      sError("%s, forward discarded, hver:%" PRIu64, pPeer->id, pHead->version);
     }
   }
 }
@@ -890,10 +890,11 @@ static int32_t syncReadPeerMsg(SSyncPeer *pPeer, SSyncHead *pHead, char *cont) {
 
   // head.len = htonl(head.len);
   if (pHead->len < 0) {
-    sError("%s, invalid pkt length, len:%d", pPeer->id, pHead->len);
+    sError("%s, invalid pkt length, hlen:%d", pPeer->id, pHead->len);
     return -1;
   }
 
+  assert(pHead->len <= TSDB_MAX_WAL_SIZE);
   int32_t bytes = taosReadMsg(pPeer->peerFd, cont, pHead->len);
   if (bytes != pHead->len) {
     sError("%s, failed to read, bytes:%d len:%d", pPeer->id, bytes, pHead->len);
diff --git a/src/sync/src/syncRetrieve.c b/src/sync/src/syncRetrieve.c
index 82a4627ea55a9f0caf041b9ee41b4b3e50c229e4..968f5becaddc7b06c06171a4d91cc0e0ffffc0da 100644
--- a/src/sync/src/syncRetrieve.c
+++ b/src/sync/src/syncRetrieve.c
@@ -244,7 +244,7 @@ static int32_t syncCheckLastWalChanges(SSyncPeer *pPeer, uint32_t *pEvent) {
 }
 
 static int32_t syncRetrieveLastWal(SSyncPeer *pPeer, char *name, uint64_t fversion, int64_t offset, uint32_t *pEvent) {
-  SWalHead *pHead = malloc(640000);
+  SWalHead *pHead = malloc(SYNC_MAX_SIZE);
   int32_t   code = -1;
   int32_t   bytes = 0;
   int32_t   sfd;
diff --git a/src/sync/src/tarbitrator.c b/src/sync/src/tarbitrator.c
index 496bf074357487ab1c87072948b2e26be056dc58..ae8f2ea518204d57891aa71ac7b4f3ee9e863ed7 100644
--- a/src/sync/src/tarbitrator.c
+++ b/src/sync/src/tarbitrator.c
@@ -86,7 +86,7 @@ int32_t main(int32_t argc, char *argv[]) {
   info.numOfThreads = 1;
   info.serverIp = 0;
   info.port = tsArbitratorPort;
-  info.bufferSize = 640000;
+  info.bufferSize = SYNC_MAX_SIZE;
   info.processBrokenLink = arbProcessBrokenLink;
   info.processIncomingMsg = arbProcessPeerMsg;
   info.processIncomingConn = arbProcessIncommingConnection;
diff --git a/src/vnode/src/vnodeMain.c b/src/vnode/src/vnodeMain.c
index bd44ce8e1cd65761a084686e4281cd237720d762..7592b4edc4c6d62c2f404242426d625d7b3518c7 100644
--- a/src/vnode/src/vnodeMain.c
+++ b/src/vnode/src/vnodeMain.c
@@ -583,6 +583,7 @@ static int vnodeProcessTsdbStatus(void *arg, int status) {
 
   if (status == TSDB_STATUS_COMMIT_OVER) {
     vDebug("vgId:%d, commit over, fver:%" PRIu64 " vver:%" PRIu64, pVnode->vgId, pVnode->fversion, pVnode->version);
+    walRemoveOldFiles(pVnode->wal);
     return vnodeSaveVersion(pVnode);
   }
 
diff --git a/src/vnode/src/vnodeWrite.c b/src/vnode/src/vnodeWrite.c
index 99594607ac0a19e31f438e0a02320ee189d55253..e67c544fb23456d8775077e8a86f1745508bd0fc 100644
--- a/src/vnode/src/vnodeWrite.c
+++ b/src/vnode/src/vnodeWrite.c
@@ -217,6 +217,11 @@ int32_t vnodeWriteToWQueue(void *vparam, void *wparam, int32_t qtype, void *rpar
     if (code != TSDB_CODE_SUCCESS) return code;
   }
 
+  if (pHead->len > TSDB_MAX_WAL_SIZE) {
+    vError("vgId:%d, wal len:%d exceeds limit, hver:%" PRIu64, pVnode->vgId, pHead->len, pHead->version);
+    return TSDB_CODE_WAL_SIZE_LIMIT;
+  }
+
   int32_t size = sizeof(SVWriteMsg) + sizeof(SWalHead) + pHead->len;
   SVWriteMsg *pWrite = taosAllocateQitem(size);
   if (pWrite == NULL) {
diff --git a/src/wal/inc/walInt.h b/src/wal/inc/walInt.h
index d1e977225929d8e34b42fbe6aec53c3f41bb7273..36311c8f5d92fe0ba97793224df59d4dd33b5da6 100644
--- a/src/wal/inc/walInt.h
+++ b/src/wal/inc/walInt.h
@@ -34,7 +34,7 @@ extern int32_t wDebugFlag;
 #define WAL_PREFIX     "wal"
 #define WAL_PREFIX_LEN 3
 #define WAL_REFRESH_MS 1000
-#define WAL_MAX_SIZE   (1024 * 1024)
+#define WAL_MAX_SIZE   (TSDB_MAX_WAL_SIZE + sizeof(SWalHead) + 16)
 #define WAL_SIGNATURE  ((uint32_t)(0xFAFBFDFE))
 #define WAL_PATH_LEN   (TSDB_FILENAME_LEN + 12)
 #define WAL_FILE_LEN   (TSDB_FILENAME_LEN + 32)
diff --git a/src/wal/src/walMgmt.c b/src/wal/src/walMgmt.c
index de666c85e832aa52ce33d442f543c0512d91adb8..9ba0dfd1240e56eec22c2f8a2eb42ec0326fc32c 100644
--- a/src/wal/src/walMgmt.c
+++ b/src/wal/src/walMgmt.c
@@ -135,7 +135,7 @@ void walClose(void *handle) {
       if (remove(pWal->name) < 0) {
         wError("vgId:%d, wal:%p file:%s, failed to remove", pWal->vgId, pWal, pWal->name);
       } else {
-        wDebug("vgId:%d, wal:%p file:%s, it is removed", pWal->vgId, pWal, pWal->name);
+        wInfo("vgId:%d, wal:%p file:%s, it is removed", pWal->vgId, pWal, pWal->name);
       }
     }
   } else {
diff --git a/src/wal/src/walWrite.c b/src/wal/src/walWrite.c
index 9681f4b898d0661d25650067362be429854e8c94..d3a41ec6b22883468625ef4a858f4d2f7d88a773 100644
--- a/src/wal/src/walWrite.c
+++ b/src/wal/src/walWrite.c
@@ -58,24 +58,32 @@ int32_t walRenew(void *handle) {
     wDebug("vgId:%d, file:%s, it is created", pWal->vgId, pWal->name);
   }
 
-  if (pWal->keep != TAOS_WAL_KEEP) {
-    // remove the oldest wal file
-    int64_t oldFileId = -1;
-    if (walGetOldFile(pWal, pWal->fileId, WAL_FILE_NUM, &oldFileId) == 0) {
-      char walName[WAL_FILE_LEN] = {0};
-      snprintf(walName, sizeof(walName), "%s/%s%" PRId64, pWal->path, WAL_PREFIX, oldFileId);
-
-      if (remove(walName) < 0) {
-        wError("vgId:%d, file:%s, failed to remove since %s", pWal->vgId, walName, strerror(errno));
-      } else {
-        wDebug("vgId:%d, file:%s, it is removed", pWal->vgId, walName);
-      }
+  pthread_mutex_unlock(&pWal->mutex);
+
+  return code;
+}
+
+void walRemoveOldFiles(void *handle) {
+  SWal *pWal = handle;
+  if (pWal == NULL) return;
+  if (pWal->keep == TAOS_WAL_KEEP) return;
+
+  pthread_mutex_lock(&pWal->mutex);
+
+  // remove the oldest wal file
+  int64_t oldFileId = -1;
+  if (walGetOldFile(pWal, pWal->fileId, WAL_FILE_NUM, &oldFileId) == 0) {
+    char walName[WAL_FILE_LEN] = {0};
+    snprintf(walName, sizeof(walName), "%s/%s%" PRId64, pWal->path, WAL_PREFIX, oldFileId);
+
+    if (remove(walName) < 0) {
+      wError("vgId:%d, file:%s, failed to remove since %s", pWal->vgId, walName, strerror(errno));
+    } else {
+      wInfo("vgId:%d, file:%s, it is removed", pWal->vgId, walName);
     }
   }
 
   pthread_mutex_unlock(&pWal->mutex);
-
-  return code;
 }
 
 int32_t walWrite(void *handle, SWalHead *pHead) {
diff --git a/tests/script/general/wal/kill.sim b/tests/script/general/wal/kill.sim
new file mode 100644
index 0000000000000000000000000000000000000000..7f103874a561c2bb2534996ab30d30ab0e8907a3
--- /dev/null
+++ b/tests/script/general/wal/kill.sim
@@ -0,0 +1,77 @@
+system sh/stop_dnodes.sh
+system sh/deploy.sh -n dnode1 -i 1
+
+print ============== deploy
+system sh/exec.sh -n dnode1 -s start 
+sleep 3001
+sql connect
+
+sql create database d1
+sql use d1
+
+sql create table t1 (ts timestamp, i int) 
+sql insert into t1 values(now, 1);
+
+print ===============  step3
+sleep 3000
+sql select * from t1;
+print rows: $rows
+if $rows != 1 then
+  return -1
+endi
+system sh/exec.sh -n dnode1 -s stop -x SIGKILL
+sleep 3000
+
+print ===============  step4
+system sh/exec.sh -n dnode1 -s start -x SIGKILL
+sleep 3000
+sql select * from t1;
+print rows: $rows
+if $rows != 1 then
+  return -1
+endi
+system sh/exec.sh -n dnode1 -s stop -x SIGKILL
+sleep 3000
+
+print ===============  step5
+system sh/exec.sh -n dnode1 -s start -x SIGKILL
+sleep 3000
+sql select * from t1;
+print rows: $rows
+if $rows != 1 then
+  return -1
+endi
+system sh/exec.sh -n dnode1 -s stop -x SIGKILL
+sleep 3000
+
+print ===============  step6
+system sh/exec.sh -n dnode1 -s start -x SIGKILL
+sleep 3000
+sql select * from t1;
+print rows: $rows
+if $rows != 1 then
+  return -1
+endi
+system sh/exec.sh -n dnode1 -s stop -x SIGKILL
+sleep 3000
+
+print ===============  step7
+system sh/exec.sh -n dnode1 -s start -x SIGKILL
+sleep 3000
+sql select * from t1;
+print rows: $rows
+if $rows != 1 then
+  return -1
+endi
+system sh/exec.sh -n dnode1 -s stop -x SIGKILL
+sleep 3000
+
+print ===============  step8
+system sh/exec.sh -n dnode1 -s start -x SIGKILL
+sleep 3000
+sql select * from t1;
+print rows: $rows
+if $rows != 1 then
+  return -1
+endi
+system sh/exec.sh -n dnode1 -s stop -x SIGKILL
diff --git a/tests/script/general/wal/sync.sim b/tests/script/general/wal/sync.sim
new file mode 100644
index 0000000000000000000000000000000000000000..abaf22f91921e5bcc8effbe6fc1c66766ff92a3f
--- /dev/null
+++ b/tests/script/general/wal/sync.sim
@@ -0,0 +1,124 @@
+system sh/stop_dnodes.sh
+
+system sh/deploy.sh -n dnode1 -i 1
+system sh/deploy.sh -n dnode2 -i 2
+system sh/deploy.sh -n dnode3 -i 3
+
+system sh/cfg.sh -n dnode1 -c numOfMnodes -v 3
+system sh/cfg.sh -n dnode2 -c numOfMnodes -v 3
+system sh/cfg.sh -n dnode3 -c numOfMnodes -v 3
+
+system sh/cfg.sh -n dnode1 -c mnodeEqualVnodeNum -v 4
+system sh/cfg.sh -n dnode2 -c mnodeEqualVnodeNum -v 4
+system sh/cfg.sh -n dnode3 -c mnodeEqualVnodeNum -v 4
+
+system sh/cfg.sh -n dnode1 -c http -v 1
+system sh/cfg.sh -n dnode2 -c http -v 1
+system sh/cfg.sh -n dnode3 -c http -v 1
+
+system sh/cfg.sh -n dnode1 -c maxTablesPerVnode -v 20000
+system sh/cfg.sh -n dnode2 -c maxTablesPerVnode -v 20000
+system sh/cfg.sh -n dnode3 -c maxTablesPerVnode -v 20000
+
+system sh/cfg.sh -n dnode1 -c replica -v 3
+system sh/cfg.sh -n dnode2 -c replica -v 3
+system sh/cfg.sh -n dnode3 -c replica -v 3
+
+system sh/cfg.sh -n dnode1 -c maxSQLLength -v 940032
+system sh/cfg.sh -n dnode2 -c maxSQLLength -v 940032
+system sh/cfg.sh -n dnode3 -c maxSQLLength -v 940032
+
+print ============== deploy
+
+system sh/exec.sh -n dnode1 -s start 
+sleep 5001
+sql connect
+
+sql create dnode $hostname2
+sql create dnode $hostname3
+system sh/exec.sh -n dnode2 -s start
+system sh/exec.sh -n dnode3 -s start 
+
+print  =============== step1
+$x = 0
+show1: 
+	$x = $x + 1
+	sleep 2000
+	if $x == 5 then
+	  return -1
+	endi
+sql show mnodes -x show1
+$mnode1Role = $data2_1
+print mnode1Role $mnode1Role
+$mnode2Role = $data2_2
+print mnode2Role $mnode2Role
+$mnode3Role = $data2_3
+print mnode3Role $mnode3Role
+
+if $mnode1Role != master then
+  goto show1
+endi
+if $mnode2Role != slave then
+  goto show1
+endi
+if $mnode3Role != slave then
+  goto show1
+endi
+
+print ===============  step2
+sql create database d1 replica 3
+sql use d1
+
+sql create table table_rest (ts timestamp, i int) 
+print sql length is 870KB
+restful d1 table_rest 1591072800 30000
+restful d1 table_rest 1591172800 30000
+restful d1 table_rest 1591272800 30000
+restful d1 table_rest 1591372800 30000
+restful d1 table_rest 1591472800 30000
+restful d1 table_rest 1591572800 30000
+restful d1 table_rest 1591672800 30000
+restful d1 table_rest 1591772800 30000
+restful d1 table_rest 1591872800 30000
+restful d1 table_rest 1591972800 30000
+
+sql select * from table_rest;
+print rows: $rows
+if $rows != 300000 then
+  return -1
+endi
+
+print ===============  step3
+system sh/exec.sh -n dnode1 -s stop -x SIGINT
+sleep 5000
+sql select * from table_rest;
+print rows: $rows
+if $rows != 300000 then
+  return -1
+endi
+system sh/exec.sh -n dnode1 -s start -x SIGINT
+sleep 5000
+
+print ===============  step4
+system sh/exec.sh -n dnode2 -s stop -x SIGINT
+sleep 5000
+sql select * from table_rest;
+print rows: $rows
+if $rows != 300000 then
+  return -1
+endi
+system sh/exec.sh -n dnode2 -s start -x SIGINT
+sleep 5000
+
+print ===============  step5
+system sh/exec.sh -n dnode3 -s stop -x SIGINT
+sleep 5000
+sql select * from table_rest;
+print rows: $rows
+if $rows != 300000 then
+  return -1
+endi
+
+system sh/exec.sh -n dnode1 -s stop  -x SIGINT
+system sh/exec.sh -n dnode2 -s stop  -x SIGINT
+system sh/exec.sh -n dnode3 -s stop  -x SIGINT
\ No newline at end of file
diff --git a/tests/script/jenkins/basic.txt b/tests/script/jenkins/basic.txt
index 1b2fe37c71f26486ba847006ecf3aa373b5f55c1..2a84172da9cbd1f72f3a1dc0de3672a82085101b 100644
--- a/tests/script/jenkins/basic.txt
+++ b/tests/script/jenkins/basic.txt
@@ -236,6 +236,9 @@ cd ../../../debug; make
 ./test.sh -f general/vector/table_query.sim
 ./test.sh -f general/vector/table_time.sim
 
+./test.sh -f general/wal/sync.sim
+./test.sh -f general/wal/kill.sim
+
 ./test.sh -f unique/account/account_create.sim
 ./test.sh -f unique/account/account_delete.sim
 ./test.sh -f unique/account/account_len.sim