Sdílet prostřednictvím

Frequently Asked Questions for Replication Administrators

  • Článek

The following questions and answers provide guidance on a variety of tasks faced by administrators of replicated databases.

Configuring Replication

Does activity need to be stopped on a database when it is published?

No. Activity can continue on a database while a publication is being created. Be aware that producing a snapshot can be resource-intensive, so it is best to generate snapshots during periods of lower activity on the database (by default a snapshot is generated when you complete the New Publication Wizard).

Are tables locked during snapshot generation?

The length of time that the locks are taken depends on the type of replication used:

  • For merge publications, the Snapshot Agent does not take any locks.
  • For transactional publications, by default the Snapshot Agent takes locks only during the initial phase of snapshot generation.
  • For snapshot publications the Snapshot Agent takes locks during the entire snapshot generation process.

Because locks prevent other users from updating the tables, the Snapshot Agent should be scheduled to execute during periods of lower activity on the database, especially for snapshot publications.

When is a subscription available; when can the subscription database be used?

A subscription is available after the snapshot has been applied to the subscription database. Even though the subscription database is accessible prior to this, the database should not be used until after the snapshot has been applied. Use Replication Monitor to check the status of snapshot generation and application:

What happens if the Snapshot Agent has not completed when the Distribution or Merge Agent starts?

It will not cause an error if the Distribution Agent or Merge Agent runs at the same time as the Snapshot Agent. However, you must be aware of the following:

Should I script my replication configuration?

Yes. Scripting the replication configuration is a key part of any disaster recovery plan for a replication topology. For more information on scripting, see Scripting Replication.

What recovery model is required on a replicated database?

Replication functions properly using any of the recovery models: simple, bulk-logged, or full. Merge replication tracks change by storing information in metadata tables. Transactional replication tracks changes by marking the transaction log, but this marking process is not affected by the recovery model.

Why does replication add a column to replicated tables; will it be removed if the table isn't published?

To track changes, merge replication and transactional replication with queued updating subscriptions must be able to uniquely identify every row in every published table. To accomplish this:

  • Merge replication adds the column rowguid to every table, unless the table already has a column of data type uniqueidentifier with the ROWGUIDCOL property set (in which case this column is used). If the table is dropped from the publication, the rowguid column is removed; if an existing column was used for tracking, the column is not removed.
  • If a transactional publication supports queued updating subscriptions, replication adds the column msrepl_tran_version to every table. If the table is dropped from the publication, the msrepl_tran_version column is not removed.

How do I manage constraints on published tables?

There are a number of issues to consider regarding constraints on published tables:

  • Transactional replication requires a primary key constraint on each published table. Merge replication does not require a primary key, but if one is present, it must be replicated. Snapshot replication does not require a primary key.
  • By default, primary key constraints, indexes, and check constraints are replicated to Subscribers.
  • The NOT FOR REPLICATION option is specified by default for foreign key constraints and check constraints; the constraints are enforced for user operations but not agent operations. For more information, see Controlling Constraints, Identities, and Triggers with NOT FOR REPLICATION.

For information on setting the schema options that control whether constraints are replicated, see How to: Specify Schema Options (SQL Server Management Studio) and How to: Specify Schema Options (Replication Transact-SQL Programming).

How do I manage identity columns?

Replication provides automatic identity range management for replication topologies that include updates at the Subscriber. For more information, see Replicating Identity Columns.

Can the same objects be published in different publications?

Yes, but with some restrictions. For more information, see the section "Publishing Tables in More Than One Publication" in the topic Publishing Data and Database Objects.

Can multiple publications use the same distribution database?

Yes. There are no restrictions on the number or types of publications that can use the same distribution database. All publications from a given Publisher must use the same Distributor and distribution database.

If you have multiple publications, you can configure multiple distribution databases at the Distributor to ensure that the data flowing through each distribution database is from a single publication. Use the Distributor Properties dialog box or sp_adddistributiondb (Transact-SQL) to add a distribution database. For more information about accessing the dialog box, see How to: View and Modify Distributor Properties (SQL Server Management Studio).

How do I find information on the Distributor and Publisher, such as which objects in a database are published?

This information is available through SQL Server Management Studio, and a number of replication stored procedures. For information, see Replication Properties and Distributor and Publisher Information Script.

Does replication encrypt data?

No. Replication does not encrypt data that is stored in the database or transferred over the network. For more information, see the "Encryption" section of the topic Elements of Replication Security.

How do I replicate data over the Internet?

Replicate data over the Internet using:

All types of Microsoft SQL Server replication can replicate data over a VPN, but you should consider Web synchronization if you are using merge replication.

Does replication resume if a connection is dropped

Yes. Replication processing resumes at the point at which it left off if a connection is dropped. If you are using merge replication over an unreliable network, consider using logical records, which ensures related changes are processed as a unit. For more information, see Grouping Changes to Related Rows with Logical Records.

Does replication work over low bandwidth connections? Does it use compression?

Yes, replication does work over low bandwidth connections. For connections over TCP/IP, it uses the compression provided by the protocol but does not provide additional compression. For Web synchronization connections over HTTPS, it uses the compression provided by the protocol and also additional compression of the XML files used to replicate changes. For more information about replicating over low bandwidth connections, see A Slow Network Is Causing Problems.

Logins and Object Ownership

Are logins and passwords replicated?

No. You could create a DTS package to transfer logins and passwords from a Publisher to one or more Subscribers. For more information, see Designing and Creating Integration Services Packages.

What are schemas and how are they replicated?

In Microsoft SQL Server 2005, schema has two meanings:

  • The definition of an object, such as a CREATE TABLE statement. By default, replication copies the definitions of all replicated objects to the Subscriber.
  • The owner of an object, which is used in a multi-part name, such as <Database>.<Schema>.<Object>. Schemas are defined using the CREATE SCHEMA statement.

Replication has the following default behavior in the New Publication Wizard with respect to schemas and object ownership:

  • For articles in merge publications with a compatibility level of 90 or higher, snapshot publications, and transactional publications: by default, the object owner at the Subscriber is the same as the owner of the corresponding object at the Publisher. If the schemas that own objects do not exist at the Subscriber, they are created automatically.
  • For articles in merge publications with a compatibility level lower than 90: by default, the owner is left blank and is specified as dbo during the creation of the object on the Subscriber.
  • For articles in Oracle publications: by default, the owner is specified as dbo.
  • For articles in publications that use character mode snapshots (which are used for non-SQL Server Subscribers and SQL Server 2005 Compact Edition Subscribers): by default, the owner is left blank. The owner defaults to the owner associated with the account used by the Distribution Agent or Merge Agent to connect to the Subscriber.

The object owner can be changed through the Article Properties - <Article> dialog box and through the following stored procedures: sp_addarticle, sp_addmergearticle, sp_changearticle, and sp_changemergearticle. For more information, see How to: View and Modify Publication and Article Properties (SQL Server Management Studio), How to: Define an Article (Replication Transact-SQL Programming), and How to: View and Modify Article Properties (Replication Transact-SQL Programming).

How can grants on the subscription database be configured to match grants on the publication database?

By default, replication does not execute GRANT statements on the subscription database. If you want the permissions on the subscription database to match those on the publication database, use one of the following methods:

What happens to permissions granted in a subscription database if a subscription is reinitialized?

By default, objects at the Subscriber are dropped and recreated when a subscription is reinitialized, which causes all granted permissions for those objects to be dropped. There are two ways to handle this:

  • Reapply the grants after the reinitialization using the techniques described in the previous section.
  • Specify that objects should not be dropped when the subscription is reinitialized. Prior to reinitialization, either:

Database Maintenance

Why can't I run TRUNCATE TABLE on a published table?

TRUNCATE TABLE is a non-logged operation that does not fire triggers. It is not permitted because replication cannot track the changes caused by the operation: transactional replication tracks changes through the transaction log; merge replication tracks changes through triggers on published tables.

What is the effect of running a bulk insert command on a replicated database?

For transactional replication, bulk inserts are tracked and replicated like other inserts. For merge replication, you must ensure that change tracking metadata is updated properly. For more information, see the section "Bulk Inserting Data into Published Tables" in Considerations for Merge Replication.

Are there any replication considerations for backup and restore?

Yes. There are a number of special considerations for databases that are involved in replication. For more information, see Backing Up and Restoring Replicated Databases.

Does replication affect the size of the transaction log?

Merge replication and snapshot replication do not affect transaction log size, but transactional replication can. If a database includes one or more transactional publications, the log is not truncated until all transactions relevant to the publications have been delivered to the distribution database. If the transaction log is growing too large, and the Log Reader Agent is running on a scheduled basis, consider shortening the interval between runs or setting it run in continuous mode. If it is set to run in continuous mode (the default), ensure that it is running. For more information on checking Log Reader Agent status, see How to: View Information and Perform Tasks for the Agents Associated With a Publication (Replication Monitor).

Additionally, if you have set the option 'sync with backup' on the publication database or distribution database, the transaction log is not truncated until all transactions have been backed up. If the transaction log is growing too large, and you have this option set, consider shortening the interval between transaction log backups. For more information on backing up and restoring databases involved in transactional replication, see Strategies for Backing Up and Restoring Snapshot and Transactional Replication.

How do I upgrade from SQL Server version 7.0 or SQL Server 2000 if I have replicated databases?

SQL Server 2005 supports upgrading replicated databases from previous versions of SQL Server. For more information, see Upgrading Replicated Databases.

How do I rebuild indexes or tables in replicated databases?

There are a variety of mechanisms for rebuilding indexes. They can all be used with no special considerations for replication, with the following exception: primary keys are required on tables in transactional publications, so you cannot drop and recreate primary keys on these tables.

How do I add or change indexes on publication and subscription databases?

Indexes can be added at the Publisher or Subscribers with no special considerations for replication (be aware that indexes can affect performance). CREATE INDEX and ALTER INDEX are not replicated, so if you add or change an index at, for example, the Publisher, you must make the same addition or change at the Subscriber if you want it reflected there.

How do I move or rename files for databases involved in replication?

In previous versions of SQL Server, moving or renaming database files required detaching and reattaching the database. Because a replicated database cannot be detached, replication had to be removed from these databases first. In SQL Server 2005, you can move or rename files without detaching and re-attaching the database, with no effect on replication. For more information about moving and renaming files, see ALTER DATABASE (Transact-SQL).

How do I drop a table that is being replicated?

First drop the article from the publication using sp_droparticle (Transact-SQL), sp_dropmergearticle (Transact-SQL), or the Publication Properties - <Publication> dialog box, and then drop it from the database using DROP <Object>. You cannot drop articles from snapshot or transactional publications after subscriptions have been added; you must drop the subscriptions first. For more information, see Adding Articles to and Dropping Articles from Existing Publications.

How do I add or drop columns on a published table?

SQL Server 2005 supports a wide variety of schema changes on published objects, including adding and dropping columns. For example, execute ALTER TABLE … DROP COLUMN at the Publisher, and the statement is replicated to Subscribers and then executed to drop the column. Subscribers running previous version of SQL Server support adding and dropping columns through the stored procedures sp_repladdcolumn (Transact-SQL) and sp_repldropcolumn (Transact-SQL). For more information, see Making Schema Changes on Publication Databases.

Replication Maintenance

How do I determine if the data at Subscribers is synchronized with data at the Publisher?

Use validation. Validation reports on whether a given Subscriber is synchronized with the Publisher. For more information, see Validating Replicated Data. Validation does not provide information on which rows if any are not synchronized correctly, but the tablediff Utility does.

How do I add a table to an existing publication?

It is not necessary to stop activity on the publication or subscription databases in order to add a table (or another object). Add a table to a publication through the Publication Properties - <Publication> dialog box or the stored procedures sp_addarticle (Transact-SQL) and sp_addmergearticle (Transact-SQL). For more information, see Adding Articles to and Dropping Articles from Existing Publications.

How do I remove a table from a publication?

Remove a table from the publication using sp_droparticle (Transact-SQL), sp_dropmergearticle (Transact-SQL), or the Publication Properties - <Publication> dialog box. You cannot drop articles from snapshot or transactional publications after subscriptions have been added; you must drop the subscriptions first. For more information, see Adding Articles to and Dropping Articles from Existing Publications.

What actions require subscriptions to be reinitialized?

There are a number of article and publication changes that require subscriptions to be reinitialized. For more information, see Changing Publication and Article Properties.

What actions cause snapshots to be invalidated?

There are a number of article and publication changes that invalidate snapshots and require a new snapshot to be generated. For more information, see Changing Publication and Article Properties.

How do I remove replication?

The actions required to remove replication from a database depend on whether the database served as a publication database, subscription database, or both. For more information, see Removing Replication.

How do I determine whether there are transactions or rows to be replicated?

For transactional replication, use stored procedures or the Undistributed Commands tab in Replication Monitor. For more information, see How to: View Replicated Commands and Other Information in the Distribution Database (Replication Transact-SQL Programming) and How to: View Information and Perform Tasks for the Agents Associated With a Subscription (Replication Monitor).

For merge replication, use the stored procedure sp_showpendingchanges. For more information, see sp_showpendingchanges (Transact-SQL).

How far behind is the Distribution Agent? Should I reinitialize?

Use the sp_replmonitorsubscriptionpendingcmds stored procedure or the Undistributed Commands tab in Replication Monitor. The stored procedure and tab display:

  • The number of commands in the distribution database that have not been delivered to the selected Subscriber. A command consists of one Transact-SQL data manipulation language (DML) statement or one data definition language (DDL) statement.
  • The estimated amount of time to deliver commands to the Subscriber. If this value is greater than the amount of time required to generate and apply a snapshot to the Subscriber, consider reinitializing the Subscriber. For more information, see Reinitializing a Subscription.

For more information, see sp_replmonitorsubscriptionpendingcmds (Transact-SQL) and How to: View Information and Perform Tasks for the Agents Associated With a Subscription (Replication Monitor).

Replication and Other Database Features

Does replication work in conjunction with log shipping and database mirroring?

Yes. For more information, see Replication and Log Shipping and Replication and Database Mirroring.

Does replication work in conjunction with clustering?

Yes. No special considerations are required because all data is stored on one set of disks on the cluster.

See Also

Concepts

Best Practices for Replication Administration

Other Resources

Administering Replication

Help and Information

Getting SQL Server 2005 Assistance