一:重启和停止服务
如许停止服务或重启服务应彻底停止服务；
启动服务顺序(服务依赖）：
1. Microsoft Exchange System Attendant
2. Microsoft Exchange Directory
3. Microsoft Exchange Information Store
4. Microsoft Exchange Messaging Transfer Agent
5. Microsoft Exchange Internet Mail Connector
6. Microsoft Exchange Key Manager

二:服务依赖关系
The DS depends on the SA
The IS depends on the DS and SA
The MTA depends on the DS and SA
The IMC depends on the SA, DS, IS, and MTA

三:内建的恢复措施
The DS, IS, and MTA all have built-in startup recovery methods
The DS and IS both go through JET recovery
If the MTA service did not stop cleanly, it will perform an MTACheck at startup

四:服务启动失败
首先检查事件察看器；
是否在修复过程中；
If the MTA will not start, try stopping the IS prior to starting the MTA. If the MTA will then start, start the IS and monitor the server
If the IMC fails to start, check the Hosts file. The IMC server must know each target server
Booting the server can do wonders. Sometimes the machine is simply in a state where a reboot is required

五:服务停止失败
If a service is in an uncontrollable state, the process will have to be terminated before the service can be brought online.
If a service stop is aborted, it will leave the service in an unknown state. Use SRVINFO and the event application log to verify the state of a service.
To force a process to terminate, use KILL.EXE or PVIEW.EXE.
Use TLIST.EXE to verify the process has been successfully terminated
Performance Monitors性能监测
MSExchangeMTA, Work Queue Length.
MSExchangeMTA, Messages Delivered per Minute.
Exchange services processes
Paging File, % Usage
LogicalDisk, Free Megabytes, Instance:
MSExchangeIS, Active Connection Count.
MSExhangeDS, Pending Replications.
MSExchangeDS, Remaining Replication Updates
Exchange Monitors
Server monitors check each listed server to verify specific services are running. A Server Monitor can be used to track any Windows NT service if desired.
Link Monitors send “ping” messages to the listed servers. If the ping is not returned in a defined amount of time, an alert is sent to the TECHNICIAN pager.
Offline Defragmentation
Offline defragmentation is performed for two reasons. The first is to reduce the size of a database. The second reason is to create a defragmented database.
Because of tree fragmentation, there may be a situation when the IS will not accept messages fast enough to keep up with the incoming volume. When this is the case, an offline compaction will create a database that is free from the problems.
Before a database is compacted, its corresponding service must be stopped. It is recommended that all of the Exchange Services be stopped. clean
Offline Defragmentation
Database Consistency ——ESEUTIL /d PRIV.EDB， Other databases are DIR.EDB and PUB.EDB
Starting and stopping the database will make the database consistent. Alternately, EDBUTIL /R /IS or EDBUTIL /R /DS will do a “soft” recovery of the database and bring it to a consistent state
Procedure
Once the Exchange services have been stopped, change to the directory that contains the database to be defragmented. It is recommended that the database be backed up before defragmentation is performed. If there is an internal problem (other than normal JET BTREE corruption), there is a chance that defragmentation will result in an unusable database.

六:Procedure
Private and Public Information Stores
1. Net Stop MSEXCHANGEIS (also stop the MTA, DS, and SA)
2. Verify the services stopped cleanly
3. Change to D:\EXCHSRVR\MDBDATA
4. (Optional) XCOPY *.* \BACKUP\xxxx\ (where xxxx is the date, for example, “0929”)
5. D:\EXCHSRVR\BIN\EDBUTIL /D PRIV.EDB /N for the Private store
6. D:\EXCHSRVR\BIN\EDBUTIL /D PUB.EDB /N for the Public store
7. Start the Exchange services
Procedure
Directory Store
1. Net Stop MSEXCHANGEDS (also stop the MTA, IS, and SA)
2. Change to D:\EXCHSRVR\DSADATA
3. (Optional) XCOPY *.* \BACKUP\xxxx\ (where xxxx is the date, for example, “0929”)
4. D:\EXCHSRVR\BIN\EDBUTIL /D DIR.EDB /N
5. Start the Exchange services

七:Outlook Web Access
Outlook Web Access describes a set of services that allows the use of a Web browser as a mail client to access various Exchange services
HTTP Settings ——The HHTP protocol is enabled in all messaging sites, at the site leve ；
IIS Settings ——Due to limitations within Windows NT/LAN Manager, these systems are configured with BASIC Authentication + SSL. An SSL certificate is required for each server to enable the service
Escalation Path
When a problem cannot be resolved by the administrator(s) on duty, it needs to be escalated to the next support group.
First Tier ——Use the others in the first-tier team to troubleshoot and solve the problem ；
Second Tier ——When first tier cannot resolve a problem, the escalation path is to the second-tier escalation team ；This team is responsible for providing a solution or escalating to MTS ；
Third-Tier Engineering ——The engineering team is responsible for providing a solution or researching the issue so it can be resolved. This team also has the responsibility of escalating to MTS if required

八:Disaster Recovery Procedures
Problems with Exchange servers generally fall into one of the categories below:
1. Hardware failures
2. Database corruption
3. Accidentally deleted mail
4. Windows NT failures
5. Drive structure failures
6 . Clone to new hardware
As long as there is a backup of the database, the server can be restored. If the Information Transaction LOG files exist, then there will be no data loss.
If there isn’t a backup, then the server must be rebuilt from scratch. The DS database is required—if it isn’t available then there is no option but to completely rebuild the server.

九:Hardware Failures
If a component is damaged, the local hardware technician can replace the component. This plan assumes that the data on the hard drives is intact.
Database Corruption
If one of the Exchange databases is corrupt, restore it from the last backup.
A corrupt database will generally be found when the Exchange backup occurs. Look for event #25. Note that the recommendation is to restore from backup.
Accidentally Deleted Mail
If a customer accidentally deletes needed mail, a second machine must be used to do the restore. Because the entire IS database must be restored in order to retrieve the mailbox, the production server can

十:Windows NT Failures
A Windows NT failure is when the operating system is corrupted and won’t start or continually crashes. Be sure that the latest service packs have been tried before assuming there is corruption. Infrequent crashes are usually related to a bug, not corruption.
Recovering from a broken operating system isn’t as difficult as it seems. Since the databases should be intact, they can be “slid” into place once the software is replaced.
The procedure is to install Windows NT and Service Packs, install a dummy version of Exchange, and then put the data back into place.
Windows NT Failures
Drive Structure Failures
This section assumes that data on the D$ drive has been lost due to massive corruption or the loss of multiple drives in the RAID5 array. Tape backups of the DS and IS databases are required.
If the LOG files aren’t available, there will be data loss. The loss will be the data since the last backup
Clone to New Hardware
In these procedures only the databases and the tracking logs files are brought forward to the new server. This renders a server with clean Windows NT and Exchange installs
PSS Troubleshooting Tips
Recovering an Information Store That Will Not Start
Some things to check first:
nSufficient disk space on the drive that holds the database and log files
Windows NT Event Log
Make a full offline backup of the Information Store. This should include all EDB and LOG files (the location of these files can be seen in the Exchange Admin program on the Database Paths page of the Server object). These files are critical to development, so they can diagnose the problem and fix the code to recover from similar situations in the future

十一:PSS Troubleshooting Tips
Recovering a directory that will not start
Some things to check first:
nSufficient disk space on the drive that holds the database and log files
Windows NT event log
Make a backup of the DSADATA subdirectory. These files are critical to development, so they can diagnose the problem and fix the code to recover from similar situations in the future.

十二:PSS Troubleshooting Tips
Recovering an MTA That Will Not Start or Terminates Unexpectedly
Some things to check first:
Sufficient disk space on the drive that holds the MTADATA subdirectory
Windows NT Event Log

十三:Collect data
The following steps will help PSS and development determine what may have caused the MTA to have problems:
n?? Stop the MTA Service.
n?? Enable Maximum MTA logging through Exchange Admin for all diagnostics except APDU (which should be set to none) and Interoperability (which should be set to medium).
n?? Start the MTA Service.
n?? Reproduce the problem.
n?? Make a full backup of the MTADATA directory, including DB*.DAT and *.LOG files.

十四:First Attempt
n?? Run MTACHECK /V /F.
?
Second Attempt (escalate):
Escalate the problem to the next support group. They will work with you to determine the best path to follow.

十五:如何防止自己的 Exchnage Server 5.5 (SP3) 中继未经授权的 SMTP 邮件。
——在 Admin 程序中，选择 IMS Connector 并选择 Properties。选择 Routing 选项卡。如果选择了 Reroute Incoming SMTP mail for other authorized domains，就会在路由框下面出现 Routing Restrictions 按钮。单击此按钮，并从下列对话框中，选择 Hosts and clients that successfully authenticate。停止和启动 IMS。

十六:如果要使寄给收件人 A 的邮件能够自动寄给收件人 B，可在服务器端执行如下操作：
用 Microsoft Exchange Administrator，单击 Recipients 按钮 (Crtl+Shift+E)
双击收件人 A
选择 Delivery Options 选项卡。在 Alternate recipient 中，单击 Modify，然后选择收件人 B。如果希望收件人 A 和收件人 B 均可收到邮件，则选中 Deliver messages to both recipient and alternate recipient。 单击 OK，应用更改。

十七:
以下是使用 Exchange 5.5 的 OWA 的一个技巧。安装 Outlook 和带 IIS 的 OWA 时，默认的 Web 页是 IIS 示例页。因而，用户要得到自己的邮件，就不得不键入 http://server.domain.com/exchange。为了更便于用户使用，我们用一个小型重新定向页，作为默认页。
对我们单位的内部人员来说，这个小东西非常好用，因为他们只需在地址栏中键入 servername，DNS 就会将其带到正确的位置。对外部人员（即，有自己 ISP 的用户），也仅需键入 servername.domain.com，而不用记住 /exchange 部分

十八:了解 Performance Optimizer
Performance Optimizer 按以下两个重要类别提供建议或进行更改：
分析并推荐 Exchange Server 数据目录的最佳硬盘位置。
分析并实现注册表中内存和其它系统资源使用情况的设置。
在下列情况下，也应在安装后手动运行 Performance Optimizer：
服务器角色变更；例如，在服务器中添加了连接器或其它功能，或从服务器中删除了连接器或其它功能。
服务器、整个站点或组织中的邮箱数显著增加或减少。
对 Exchange Server 计算机进行了硬件更改，如添加 RAM 或更改硬盘配置。

十九:了解 Performance Optimizer
在更改 Performance Optimizer 设置时，请牢记下列事项：
手动进行过大更改可能导致系统不稳定。应以较小增量进行更改（不超过当前值的 10% 到 20%）。
一次解决一个问题，并记录每个更改在几天时间内对服务器稳定性和性能的影响。尽管有时更改一个参数可能需要同时更改另外一个参数，但一次致力于解决一个问题的一般原则仍适用。
有关您要进行的更改将产生的结果，可请教 Microsoft Technical Supppot。Microsoft Knowledge Base 也包含有关已知能解决特定问题的 Performance Optimizer 更改的文章。可以搜索接收到的错误，或感兴趣的 Performance Optimizer 参数。根据臆测进行手动更改可能引发问题，因此不建议这样做。
对 Performance Optimizer 设置进行小的更改不是解决硬件不足的有效方案。减小某方面的参数值以提高其它方面的性能未必有效。Performance Optimizer 的内置选择可以很好地调整硬件配置。如果站点中的拓扑和消息传递模式非常与众不同，进行手动更改也许会有帮助。
请只为解决特定问题而更改 Performance Optimizer 设置，不要试图进行全面更改。如果系统实际上没有什么问题，在您手动更改了 Performance Optimizer 中的一半参数以后，可能就有问题了。
配置，管理并使用公共文件夹
在配置之前开发一个公共文件夹策略。
定义顶层文件夹创建以避免太多的根文件夹。
利用管理员程序将容许和其他属性传播到大的文件夹数。
通过使用服务器站点，复制，和亲缘关系，将注意离集中在客户连通性上。
为了完成客户访问，回顾域信任。
配置公共文件夹，并考虑向用户提供创建协作解决方案的能力。

二十:公共文件夹使用技巧
加入定制表单来提供或检查信息。利用Design Outlook Forms特性，在Outlook中构建定制表单，对于一般用户来说是很自然很容易的事。
定义浏览来组织和发现信息。浏览可以是信息共享更容易。浏览允许对缺省选项进行个人化和简化处理以便使公共文件夹中的信息更加有吸引力，并且容易阅读。不只是提供文档和发送者的列表，文档作者及其最后编辑时间也被显示，在Office文档中定义的定制Office文档属性可以被升级。
创建规则来处理新的信息。
分配个人访问容许。

二十一:增强 Exchange 的安全性

Microsoft Windows NT；
微软服务包；
加密
病毒保护
服务账号
Internet连接
代理服务器
Microsoft Exchange中的安全性设置

原文转自：http://www.ltesting.net