全备份失败后，如何手工清除exchange日志文件，附微软KB

oneway1

Exchange如果不采用循环日志，在成功实现全备份前，将产生大量DB日志。当磁盘性能比较差时，Exchange将延缓日志提交到数据库。因此这部分日志绝对不能删除，否则DB会有问题。而那些已经提交的db的log，其实可以手工delete掉（普通删除文件的方法）。当备份出现问题，磁盘容量又不够时，该方法非常有用。

以下KB就是帮助用户如何确定哪些日志已经提交，可以删除。

XADM: Using Eseutil to Determine Which Logs Have Been Committed
View products that this article applies to.
Article ID : 182961
Last Review : October 28, 2006
Revision : 4.3
This article was previously published under Q182961
On This Page

SUMMARY

MORE INFORMATION

Exchange Server 5.5

Exchange 2000 Server
SUMMARY
You can use the Eseutil utility to view the contents of the checkpoint file to determine which log files have been committed to an Exchange database.
Back to the top

MORE INFORMATION
Exchange Server 5.5
The information in this section applies to all three Exchange Server 5.5 databases: • Dir.edb
• Priv.edb
• Pub.edb
To display the Edb.chk (checkpoint) file on the screen, run the following command from a command prompt:
eseutil /mk edb.chk | more
If you do not run eseutil from the folder where the Edb.chk file is located, you must provide the full path to the file.

In Exchange Server 5.5, there is one checkpoint file for each set of log files. There will be one Edb.chk file for the directory and one for the information store. (Priv.edb and Pub.edb share the same log files and checkpoint file.)

The information that is displayed is similar to:    Microsoft(R) Exchange Server Database Utilities

   Version 5.5

   Copyright (C) Microsoft Corporation 1991-1997. All Rights Reserved.



   Initiating FILE DUMP mode...
      Checkpoint file: edb.chk

      LastFullBackupCheckpoint (8,615,203)
      Checkpoint (22,6485,399)
      FullBackup (8,615,203)
      FullBackup time:3/6/1998 11:52:16
      IncBackup (0,0,0)
      IncBackup time:0/0/1900 0:0:0
      Signature: Create time:2/24/1998 14:23:47 Rand:24842 Computer:
      Env (Session, Opentbl, VerPage, Cursors, LogBufs, LogFile, Buffers)
          (    168,   25200,    4440,    8400,      84,   10240,    8045)

   Operation completed successfully in 0.20 seconds.
                               
Take the first number in the parentheses for the Checkpoint field and convert it to hexadecimal. The hexadecimal number is the file name of the first uncommitted log for the appropriate directory. Circular logging works by deleting all log files with generations that are lower than the generation in the checkpoint, for example:
Checkpoint (22,6485,399) = EDB00016.LOG
This syntax applies for most other fields in the dump file (LastFullBackupCheckpoint, FullBackup, IncBackup, and other fields).
Back to the top

Exchange 2000 Server
In Exchange 2000, there is one checkpoint file for each storage group and one for the directory. When you try to identify the last committed log file for a storage group, note that the storage group prefix applies to the checkpoint file and to all of the log files. For example, the default first storage group's checkpoint file name is E00.chk, and its log files are E00xxxxxx.log (where xxxxx is the hexadecimal sequence number of the log file).

To display the Exx.chk (checkpoint) file on the screen, run the following command from a command prompt (from the Exchsrvr\Bin folder):
eseutil /mk the_full_path_to_the_checkpoint_file
For example, you may run the following command:
eseutil /mk "C:\Program Files\Exchsrvr\MDBDATA\E00.chk"
Note that you must use the quotation marks if there is a space in the path to the checkpoint file.

The output of this command is similar to:    Microsoft(R) Exchange Server(TM) Database Utilities
   Version 6.0
   Copyright (C) Microsoft Corporation 1991-2000.  All Rights Reserved.

   Initiating FILE DUMP mode...
         Checkpoint file: C:\Program Files\Exchsrvr\MDBDATA\E00.chk

         LastFullBackupCheckpoint: (0x0,0,0)
         Checkpoint: (0x6A,1119,3D)
         FullBackup: (0x0,0,0)
         FullBackup time: 00/00/1900 00:00:00
         IncBackup: (0x0,0,0)
         IncBackup time: 00/00/1900 00:00:00
         Signature: Create time:09/24/2001 17:10:26 Rand:522553071 Computer:
         Env (CircLog,Session,Opentbl,VerPage,Cursors,LogBufs,LogFile,Buffers)
             (    off,    202,  30300,   1365,  10100,    128,  10240,  97940)

   Operation completed successfully in 1.192 seconds.
                               
Take the first number in the parentheses for the Checkpoint field. The hexadecimal number is the file name of the first uncommitted log for the appropriate database; in this example:
Checkpoint (0x6A,1119,3D) = E000006A.log
Back to the top

oneway1

补充，exchange 2003和2000都适用。删除日志文件请千万小心。

还有种方法，就是先将日志设成循环日志，exchange会强制提交所有日志，同时会删除没用的日志文件。不过记得好像要重启service。

[ 本帖最后由 oneway1 于 2006-12-29 09:10 编辑 ]

砸锅卖铁

实用

iwill_mail

原帖由 oneway1 于 2006-12-29 09:06 发表
补充，exchange 2003和2000都适用。删除日志文件请千万小心。

还有种方法，就是先将日志设成循环日志，exchange会强制提交所有日志，同时会删除没用的日志文件。不过记得好像要重启service。

这个方法行不行啊? 我现在就遇到这种问题,  完全备份后,日志没有自动删除, 日志太多了,但还是不敢手工删除.

oneway1

两种方法我全做过，我这儿1分钟生成一个5M log。 备份有问题的话，一下就很多了。

第一种需要看仔细，第二种简单。

不过第二种可能需要点时间，因为系统删日志需要时间。

oneway1

全备份后没有自动删除日志，需要检查全备份是否有问题。

如果全备有问题，等到想恢复的时候会有种吃苍蝇的感觉。

oneway1

全备份后没有自动删除日志，需要检查全备份是否有问题。

如果全备有问题，等到想恢复的时候会有种吃苍蝇的感觉。

砸锅卖铁

需要全备份整个存储组

oneway1

对，log是基于储存组。

iwill_mail

有没有很好的办法解决全备份不能删除日志的问题？