事件探查器捕获的sql语句文件，用Read80Trace分析，出现错误帮个忙第一回用

[fscnt]

01/21/08 11:30:59 [2008]-----------------------------------------------------

01/21/08 11:30:59 [2008]Read80Trace version 8.10.0010

01/21/08 11:30:59 [2008]-f

01/21/08 11:30:59 [2008]-drml

01/21/08 11:30:59 [2008]-II:\rml\ss.trc

01/21/08 11:30:59 [2008]Establishing database connections:

01/21/08 11:30:59 [2008]Server: (local)

01/21/08 11:30:59 [2008]Database: rml

01/21/08 11:30:59 [2008]Authentication: Windows

01/21/08 11:31:06 [2008]Processing: I:\rml\ss.trc

01/21/08 11:31:06 [2008]ERROR: Event 10 does not contain the following required column(s): 15

01/21/08 11:31:06 [2008]WARNING: Event 10 does not contain the following recommended column(s): 3

01/21/08 11:31:06 [2008]WARNING: Event 11 does not contain the following recommended column(s): 3

01/21/08 11:31:06 [2008]ERROR: Event 12 does not contain the following required column(s): 15

01/21/08 11:31:06 [2008]WARNING: Event 12 does not contain the following recommended column(s): 3

01/21/08 11:31:06 [2008]WARNING: Event 13 does not contain the following recommended column(s): 3

01/21/08 11:31:06 [2008]ERROR: Event 14 does not contain the following required column(s): 15

01/21/08 11:31:06 [2008]ERROR: Event 15 does not contain the following required column(s): 15

01/21/08 11:31:06 [2008]ERROR: Event 17 does not contain the following required column(s): 15

01/21/08 11:31:06 [2008]WARNING: Statement level analysis is disabled because the SP:StmtStarting/SP:StmtCompleted event(s) do not contain the following required columns: 22, 29

01/21/08 11:31:06 [2008]WARNING: Event 45 does not contain the following recommended column(s): 3, 25

01/21/08 11:31:06 [2008]ERROR: Performance analysis has been disabled because one or more events is missing required columns. Consult the help file for the recommended set of trace events and necessary columns.

01/21/08 11:31:06 [2008]Reads completed - Global Error Status 120

01/21/08 11:31:06 [2008]READER: Total Read Count: 5 Read Stall Count: 1 accounting for 78 ms in reader thread

01/21/08 11:31:06 [2008]Signaling all workers to complete

01/21/08 11:31:06 [2008]Handling final IOs and closing open files

01/21/08 11:31:06 [2008]Flushing and Compressing SPID Files

01/21/08 11:31:06 [2008]Total Events Processed: 0

01/21/08 11:31:06 [2008]Memory Pressure Intervals Encountered: 2

[fscnt]

不明白呀

[yangjiexi]

帮顶一下!!!

[-狙击手-]

列有问题？

[fscnt]

哪列呢？

[qiule]

顶

[-狙击手-]

你没说明用的是哪个表，表的结构　什么，无从说起

[dawugui]

关注,帮顶.

[w2jc]

你的trace文件是用什么生成的？
追踪的数据库的版本是2000， 2005， 2008？

[fscnt]

http://www.microsoft.com/china/msdn/library/data/sqlserver/Profiler.mspx?mfr=true
我用的是这里的方法。
用户的数据库

[fscnt]

2000

[fscnt]

ostress.log 里边的日志是
Unable to retrieve current database context from login message
我是这么设置的
TSQL:
exec Propared SQL
SQL:BatchCompleted
SQL:BatchStarting
安全审核:
Audit login
Audit Logout
存储过程:
RPC:Completed
RPC:Starting
SP:Completed
SP:StmtCompleted
SP:StmtStarting
会话:
ExistingConnection
性能:
Show Plan statistics


-->数据列:
确认包括：
DatabaseID
EndTime
ObjectID
NestLevel
IntegerData
EventClass
TextData
ApplicationName
NTUserName
Loginname
CPU
Reads
Writes
Duration
ClientProcessID
SPID
StartTime

[w2jc]

看你的profiler的设置看不出什么问题，
不过，我想起来，如果你要重放的话，里面专门有个模板是“重放”或者“replay”，
你可以试试用这个模板，然后再运行ostress。

另外，如果你有64个以上的数据库，ostress通常会出错（但是这不太可能吧？: )

参考一下MS的：
描述 Read80Trace 和 OSTRESS 分析 SQLServer 性能实用工具
http://support.microsoft.com/kb/887057/zh-cn

[fscnt]

真是超过64个库了。
太感谢了。谢谢呀

[fscnt]

问题解决了。是数据列问题 还有点问题运行中出现OSTRESS.exe 错误

这是错误日志 Unable to retrieve current database context from login message


01/22/08 10:18:47 [288]-----------------------------------------------------

01/22/08 10:18:47 [288]Read80Trace version 8.10.0010

01/22/08 10:18:47 [288]-f

01/22/08 10:18:47 [288]-drml

01/22/08 10:18:47 [288]-Ii:\rml\ss.trc

01/22/08 10:18:47 [288]Establishing database connections:

01/22/08 10:18:47 [288]Server: (local)

01/22/08 10:18:47 [288]Database: rml

01/22/08 10:18:47 [288]Authentication: Windows

01/22/08 10:18:52 [288]Processing: i:\rml\ss.trc

01/22/08 10:18:53 [288]Reads completed - Global Error Status 0

01/22/08 10:18:56 [288]READER: Total Read Count: 19 Read Stall Count: 1 accounting for 31 ms in reader thread

01/22/08 10:18:56 [288]Signaling all workers to complete

01/22/08 10:18:56 [288]Handling final IOs and closing open files

01/22/08 10:18:56 [288]Flushing and Compressing SPID Files

01/22/08 10:18:56 [288]Total Events Processed: 6744

01/22/08 10:18:56 [288]Memory Pressure Intervals Encountered: 5

01/22/08 10:18:56 [288]Performance analysis hash table sizes (bytes): Batch=64532, Stmt=0, Plan=0, Proc=0

01/22/08 10:18:57 [288]Table dbo.tblUniqueBatches: loaded ~278 rows

01/22/08 10:18:57 [288]Table dbo.tblUniqueStatements: loaded ~0 rows

01/22/08 10:18:57 [288]Table dbo.tblUniquePlans: loaded ~0 rows

01/22/08 10:18:57 [288]Table dbo.tblUniquePlanRows: loaded ~0 rows

01/22/08 10:18:57 [288]Table dbo.tblBatches: loaded ~1555 rows

01/22/08 10:18:57 [288]Table dbo.tblStatements: loaded ~0 rows

01/22/08 10:18:57 [288]Table dbo.tblPlans: loaded ~440 rows

01/22/08 10:18:57 [288]Table dbo.tblPlanRows: loaded ~0 rows

01/22/08 10:18:57 [288]Table dbo.tblInterestingEvents: loaded ~0 rows

01/22/08 10:18:57 [288]Table dbo.tblConnections: loaded ~70 rows

01/22/08 10:18:57 [288]Indexing tables...

01/22/08 10:18:58 [288]Doing post-load data cleanup...

01/22/08 10:18:58 [288]Computing partial aggregates...

01/22/08 10:18:58 [288]Exporting analysis data...

01/22/08 10:19:02 [288]OSTRESS returned an error code of -1073741819. See ostress.log for the error details.

01/22/08 10:19:02 [288]Export of analysis data failed but all data is committed to the database. If OSTRESS returned a failure the data can be exported by running it from the command line; see the help file for details.

01/22/08 10:19:02 [288]Data load completed.

01/22/08 10:19:02 [288]Performance analysis failed to complete all necessary post-load processing.

[w2jc]

2000的数据库用read80trace是可以的。
出现这些问题可能是因为你用profiler的时候，里面有些事件没有包含在里面，
所以等到read80trace去读的时候就找不到这些事件对应的列。

看看日志文件 ostress.log 里面，也许会有提示。
如果要重放你的追踪文件，你的profiler模板里面至少要包含“开始时间”这个事件，
不太确定，你看看 ostress 的文档。

重放trace文件据说是比较难搞，就连微软技术支持的人都尽量不用，算是实在不得已才用。
OSTRESS.exe 一般是把那些大查询提出来，放一个文件里，用 ostress 去运行，看服务器的运行，
直接用trace文件的话，没做过，不清楚了....

[sheepcyk]

mark