MS SQL 日常维护管理常用脚本(二)

监控数据库运行

下面是整理、收集监控数据库运行的一些常用脚本，也是MS SQL 日常维护管理常用脚本(一)的续集，欢迎大家补充、提意见。

查看数据库登录名信息

Code Snippet

1. SELECT name                                AS LoginName ,
2. dbname                              AS DefaultDB ,
3. createdate                          AS CreateDate,
4. updatedate                          AS UpdateDate,
5. language                            AS Language  ,
6. CASE WHEN isntname = 1 THEN ‘NT USER’
7. ELSE ‘SQL USER’    END          AS UserType
8. FROM syslogins;

查看数据库用户信息

SELECT * FROM sysusers;

查看用户拥有的服务器角色

方法1： 用SSMS管理工具查看

方法2： 脚本查询

查看用户角色

1. SELECT name            ,
2. CASE WHEN sysadmin     = 1       THEN ‘yes’     ELSE ” END AS IsSysadmin        ,
3. CASE WHEN dbcreator    = 1       THEN ‘yes’     ELSE ” END AS IsDbCreate        ,
4. CASE WHEN securityadmin= 1       THEN ‘yes’     ELSE ” END AS IsSecurityadmin   ,
5. CASE WHEN bulkadmin    = 1       THEN ‘yes’     ELSE ” END AS IsBulkadmin       ,
6. CASE WHEN diskadmin    = 1       THEN ‘yes’     ELSE ” END AS IsDiskadmin       ,
7. CASE WHEN processadmin = 1       THEN ‘yes’     ELSE ” END AS IsProcessadmin    ,
8. CASE WHEN serveradmin  = 1       THEN ‘yes’     ELSE ” END AS IsServeradmin     ,
9. CASE WHEN setupadmin   = 1       THEN ‘yes’     ELSE ” END AS IsSetupadmin
10. FROM syslogins
11. –WHERE NAME=’loginname’

查看最大工作线程数

Code Snippet

1. SELECT  max_workers_count
2. FROM  sys.dm_os_sys_info

查看当前用户进程的会话ID

SELECT @@SPID

查询当前会话使用哪种协议

Code Snippet

1. SELECT net_transport
2. FROM   sys.dm_exec_connections
3. WHERE session_id = @@SPID;

查看当前连接的会话信息

–进程号1–50是SQL Server系统内部用的

SELECT * FROM sys.dm_exec_sessions WHERE session_id >=51

–查看某台机器的连接会话信息

SELECT * FROM sys.dm_exec_sessions WHERE session_id >=51 AND host_name='PO130018801'

–查看某个登录名的连接会话信息

SELECT * FROM sys.dm_exec_sessions WHERE session_id >=51 AND login_name='username'

–查看活动的连接会话信息

SELECT * FROM sys.dm_exec_sessions WITH(NOWAIT) WHERE session_id >=51 AND status ='running'

–查找连接到服务器的用户并返回每个用户的会话数

SELECT  login_name ,

        COUNT(session_id) AS session_count

FROM    sys.dm_exec_sessions

GROUP BY login_name ;

<!–
.csharpcode, .csharpcode pre
{
font-size: small;
color: black;
font-family: consolas, “Courier New”, courier, monospace;
background-color: #ffffff;
/*white-space: pre;*/
}
.csharpcode pre { margin: 0em; }
.csharpcode .rem { color: #008000; }
.csharpcode .kwrd { color: #0000ff; }
.csharpcode .str { color: #006080; }
.csharpcode .op { color: #0000c0; }
.csharpcode .preproc { color: #cc6633; }
.csharpcode .asp { background-color: #ffff00; }
.csharpcode .html { color: #800000; }
.csharpcode .attr { color: #ff0000; }
.csharpcode .alt
{
background-color: #f4f4f4;
width: 100%;
margin: 0em;
}
.csharpcode .lnum { color: #606060; }
–>

查看正在执行的SQL语句

方法1： 选择数据库实例，单击右键，选择”活动监视器“，监控/查看正在执行的SQL

方法2： profile去跟踪，比较耗费资源。

方法3：

Code Snippet

1. SELECT[Spid] = session_Id ,
2. ecid ,
3. [Database] = DB_NAME(sp.dbid) ,
4. [User] = nt_username ,
5. [Status] = er.status ,
6. [Wait] = wait_type ,
7. [Individual Query] = SUBSTRING(qt.text, er.statement_start_offset / 2,
8. ( CASE WHEN er.statement_end_offset = -1
9. THEN LEN(CONVERT(NVARCHAR(MAX), qt.text))
10. * 2
11. ELSE er.statement_end_offset
12. END – er.statement_start_offset ) / 2) ,
13. [Parent Query] = qt.text ,
14. Program = program_name ,
15. Hostname ,
16. nt_domain ,
17. start_time
18. FROMsys.dm_exec_requests er
19. INNER JOIN sys.sysprocesses sp ON er.session_id = sp.spid
20. CROSS APPLY sys.dm_exec_sql_text(er.sql_handle) AS qt
21. WHEREsession_Id >= 51

方法4：

Code Snippet

1. SELECT   m.session_id ,
2. m.start_time ,
3. m.command    ,
4. m.wait_type  ,
5. m.cpu_time   ,
6. CAST(s.text AS VARCHAR(1000)) AS sqlText
7. FROMmaster.sys.dm_exec_requests m WITH ( NOLOCK )
8. CROSS APPLY fn_get_sql(m.sql_handle) s
9. SELECT  r.session_id,
10. r.start_time      ,
11. r.command         ,
12. r.wait_type       ,
13. r.cpu_time        ,
14. s.text
15. FROMsys.dm_exec_requests r
16. CROSS APPLY sys.dm_exec_sql_text(r.sql_handle) s

–查看某个会话ID正在执行的SQL

Code Snippet

1. SELECT  m.session_id ,
2. m.start_time ,
3. m.command    ,
4. m.wait_type  ,
5. m.cpu_time   ,
6. CAST(s.text AS VARCHAR(1000)) AS sqlText
7. FROMmaster.sys.dm_exec_requests m WITH ( NOLOCK )
8. CROSS APPLY fn_get_sql(m.sql_handle) s
9. WHEREm.session_id = 342
10. SELECT   r.session_id      ,
11. r.start_time      ,
12. r.command         ,
13. r.wait_type       ,
14. r.cpu_time        ,
15. s.text
16. FROM sys.dm_exec_requests r
17. CROSS APPLY sys.dm_exec_sql_text(r.sql_handle) s
18. WHERE r.seesion_id =342

查看SQL SERVER进程执行的语句

Code Snippet

1. USE master
2. DECLARE @spid INT ;
3. DECLARE @sql_handle BINARY(20) ;
4. SET @spid = 56
5. SELECT@sql_handle = sql_handle
6. FROMsysprocesses AS A WITH ( NOLOCK )
7. WHEREspid = @spid ;
8. SELECTtext
9. FROM::fn_get_sql(@sql_handle) ;

查找TOP N语句

按平均 CPU 时间返回排名前十个的查询的相关信息。此示例将根据查询的查询哈希对查询进行聚合，以便按照查询的累积资源消耗来分组在逻辑上等效的查询。
–注意：SQL 2005 某些版本，没有sys.dm_exec_query_stats系统动态视图没有query_hash视图。

Code Snippet

1. USE DBNAME;
2. GO
3. SELECT TOP 10 query_stats.query_hash             AS “Query Hash”,
4. SUM(query_stats.total_worker_time) /
5. SUM(query_stats.execution_count)             AS “Avg CPU Time”,
6. MIN(query_stats.statement_text)              AS “Statement Text”
7. FROM
8. (SELECT QS.*,
9. SUBSTRING(ST.text,(QS.statement_start_offset/2) + 1,
10. ((CASE statement_end_offset
11. WHEN -1 THEN DATALENGTH(st.text)
12. ELSE QS.statement_end_offset END
13. – QS.statement_start_offset)/2) + 1) AS statement_text
14. FROM sys.dm_exec_query_stats AS QS
15. CROSS APPLY sys.dm_exec_sql_text(QS.sql_handle) as ST) as query_stats
16. GROUP BY query_stats.query_hash
17. ORDER BY 2 DESC;
18. GO

查看会话阻塞/死锁信息

方法1：查看那个引起阻塞，查看blk不为0的记录，如果存在阻塞进程，则是该阻塞进程的会话 ID。否则该列为零。 

EXEC sp_who active

方法2：查看那个引起阻塞，查看字段BlkBy，这个能够得到比sp_who更多的信息。

EXEC sp_who2 active

方法3：sp_lock 系统存储过程，报告有关锁的信息，但是不方便定位问题

方法4：sp_who_lock存储过程

Code Snippet

1. USE master;
2. GO
3. SET ANSI_NULLS ON;
4. GO
5. SET QUOTED_IDENTIFIER ON;
6. GO
7. IF EXISTS(SELECT 1 FROM sysobjects WHERE id=OBJECT_ID(N’sp_who_lock’)
8. AND OBJECTPROPERTY(id, ‘IsProcedure’) =1)
9. DROP PROCEDURE sp_who_lock;
10. GO
11. –==================================================================================================
12. —            ProcedureName             :            sp_who_lock
13. —            Author                    :            作者不详,出自网络
14. —            CreateDate                :            2013-05-13
15. —            Description               :            查看阻塞和死锁信息
16. /**************************************************************************************************
17. Parameters                    :              参数说明
18. ***************************************************************************************************
19. 无参存储过程
20. ***************************************************************************************************
21. Modified Date            Modified User     Version                 Modified Reason
22. ***************************************************************************************************
23. 2013-06-03                    Kerry        V01.00.01 调整存储过程格式,代码部分修改以及增加注释信息
24. ***************************************************************************************************/
25. –=================================================================================================
26. CREATE PROCEDURE sp_who_lock
27. AS
28. BEGIN
29. DECLARE @spid                            INT;
30. DECLARE @block                           INT;
31. DECLARE @RowCount                        INT;
32. DECLARE @RowIndex                        INT;
33. –创建临时表，保持被阻塞或正阻塞其他SQL的SQL语句信息
34. CREATE TABLE #tmp_lock_who
35. (
36. id    INT IDENTITY(1, 1) ,
37. spid    SMALLINT ,
38. block SMALLINT
39. )
40. IF @@ERROR<>0 RETURN @@ERROR;
41. INSERT INTO #tmp_lock_who
42. (
43. spid ,
44. block
45. )
46. SELECT  0 ,
47. blocked
48. FROM    ( SELECT    *
49. FROM      sysprocesses
50. WHERE     blocked > 0
51. ) a
52. WHERE   NOT EXISTS( SELECT *
53. FROM   ( SELECT    *
54. FROM      sysprocesses
55. WHERE     blocked > 0
56. ) b
57. WHERE  a.blocked = spid )
58. UNION
59. SELECT  spid ,
60. blocked
61. FROM    sysprocesses
62. WHERE   blocked > 0;
63. IF @@ERROR<>0 RETURN @@ERROR;
64. — 找到临时表的记录数
65. SELECT@RowCount = COUNT(1) ,
66. @RowIndex = 1
67. FROM#tmp_lock_who
68. IF @@ERROR<>0 RETURN @@ERROR;
69. IF@RowCount=0
70. SELECT  N’现在没有阻塞和死锁信息’ AS MESSAGE;
71. — — 循环开始
72. WHILE @RowIndex <= @RowCount
73. BEGIN
74. — 取第一条记录
75. SELECT     @spid    = spid,
76. @block   = block
77. FROM #tmp_lock_who
78. WHERE Id = @RowIndex
79. IF @spid = 0
80. SELECT N’引起数据库死锁的是: ‘ + CAST(@block AS VARCHAR(10))
81. +  N’进程号,其执行的SQL语法如下’;
82. ELSE
83. SELECT N’进程号SPID：’ + CAST(@spid AS VARCHAR(10)) + N’被进程号SPID：’
84. +  CAST(@block AS VARCHAR(10)) +’阻塞,其当前进程执行的SQL语法如下’;
85. DBCC INPUTBUFFER(@block )
86. SET @RowIndex = @RowIndex + 1;
87. END;
88. DROP TABLE #tmp_lock_who;
89. RETURN 0;
90. END

方法5：右键服务器-选择“活动和监视器”，查看进程选项。注意“任务状态”字段。

方法6：右键服务名称-选择报表-标准报表-活动-所有正在阻塞的事务。

小结：总结之后，才发现居然有这么多方法，MGD，系统的整理、梳理知识点是非常有必要的，你能更全面、深入的了解。

查看内存状态

dbcc memorystatus

具体如何分析，请查看官方文档http://support.microsoft.com/kb/907877/zh-cn

查看脚本执行时间

方法1： 查看SSMS管理器，查询窗口右下角
方法2:

Code Snippet

1. DECLARE @exectime DATETIME
2. SELECT@exectime = GETDATE()
3. –SQL 语句
4. PRINT N’SQL执行耗时:’ + CONVERT(VARCHAR(10), DATEDIFF(ms, @exectime, GETDATE()))

方法3：

SET STATISTICS TIME ON

–SQL 语句

查看进程正在执行的SQL语句 

　　dbcc inputbuffer ()

查看那些表缺少索引 

下面语句功能强大，执行结果受统计信息的影响

Code Snippet

1. SELECT sys.objects.name table_name,
2. mid.statement full_name,
3. (migs.avg_total_user_cost * migs.avg_user_impact) *(migs.user_seeks + migs.user_scans) AS Impact,
4. migs.avg_user_impact *(migs.user_seeks + migs.user_scans) Avg_Estimated_Impact,
5. ‘CREATE NONCLUSTERED INDEX IDX_’ + sys.objects.name + ‘_N ON ‘
6. + sys.objects.name COLLATE DATABASE_DEFAULT
7. + ‘ ( ‘ + IsNull(mid.equality_columns, ”)
8. + CASE WHEN mid.inequality_columns IS NULL
9. THEN ”
10. ELSE
11. CASE WHEN mid.equality_columns IS NULL
12. THEN ”
13. ELSE ‘,’
14. END + mid.inequality_columns
15. END + ‘ ) ‘
16. + CASE WHEN mid.included_columns IS NULL
17. THEN ”
18. ELSE ‘INCLUDE (‘ + mid.included_columns + ‘)’ END
19. + ‘;’ AS CreateIndexStatement,
20. mid.equality_columns,
21. mid.inequality_columns,
22. mid.included_columns
23. FROM sys.dm_db_missing_index_group_stats AS migs
24. INNER JOIN sys.dm_db_missing_index_groups AS mig ON migs.group_handle = mig.index_group_handle
25. INNER JOIN sys.dm_db_missing_index_details AS mid ON mig.index_handle = mid.index_handle
26. AND mid.database_id = DB_ID()
27. INNER JOIN sys.objects WITH (nolock) ON mid.OBJECT_ID = sys.objects.OBJECT_ID
28. WHERE (migs.group_handle IN
29. (
30. SELECT TOP (500) group_handle
31. FROM sys.dm_db_missing_index_group_stats WITH (nolock)
32. ORDER BY (avg_total_user_cost * avg_user_impact) *(user_seeks + user_scans) DESC))
33. AND OBJECTPROPERTY(sys.objects.OBJECT_ID, ‘isusertable’)=1
34. –ORDER BY [Impact] DESC, [full_name] DESC
35. ORDER BY [table_name], [Impact] desc

查看应该被移除的索引

查看那些多余的、应该被移除的索引

SQL 1：

Code Snippet

1. SELECT OBJECTNAME = OBJECT_NAME(I.OBJECT_ID),
2. INDEXNAME = I.NAME,
3. I.INDEX_ID
4. FROM SYS.INDEXES I
5. JOIN SYS.OBJECTS O ON I.OBJECT_ID = O.OBJECT_ID
6. WHERE OBJECTPROPERTY(O.OBJECT_ID,’IsUserTable’) = 1
7. AND I.INDEX_ID NOT IN(
8. SELECT S.INDEX_ID
9. FROM SYS.DM_DB_INDEX_USAGE_STATS S
10. WHERE S.OBJECT_ID = I.OBJECT_ID
11. AND I.INDEX_ID = S.INDEX_ID
12. AND DATABASE_ID = DB_ID())
13. ORDER BY OBJECTNAME, I.INDEX_ID, INDEXNAME ASC

SQL 2: 分析下面语句，移除那些没有必要的索引

Code Snippet

1. SELECT DB_NAME(database_id)         AS  N’DataBaseName’  ,
2. OBJECT_NAME(U.object_id)     AS  N’Table_Name’    ,
3. I.name                       AS  N’Index_Name’    ,
4. user_seeks                   AS  N’用户索引查找次数’,
5. user_scans                   AS  N’用户索引扫描次数’,
6. last_user_seek               AS  N’最后查找时间’   ,
7. last_user_scan               AS  N’最后扫描时间’   ,
8. rows                         AS  N’表中的行数’
9. FROM sys.dm_db_index_usage_stats AS U
10. INNER JOIN sys.indexes I ON U.index_id= I.index_idAND U.object_id= I.object_id
11. INNER JOIN sysindexesT ON I.object_id = T.id
12. WHERE database_id= DB_ID(‘DbName’)
13. AND OBJECT_NAME(U.object_id)=’TableName’
14. ORDER BY user_seeks, user_scans, object_name(U.object_id);