Even with this trace flag enabled, the log reader always waits for the synchronous secondary replicas(They could become asynchronous commit mode as documented here, so log reader can move forward.). Trace flag 1448 enables the replication log reader to move forward even if the asynchronous secondary replicas have not acknowledged the reception of a change. Then add a new secondary replica to the availability group. To allow the log reader to proceed further and still have disaster recovery capacity, remove the original primary replica from the availability group using ALTER AVAILABITY GROUP REMOVE REPLICA. The log reader, now running against the secondary database, will not proceed forward since Always On cannot harden any changes to any secondary database. When the publisher has only two availability replicas (one primary and one secondary) and a failover happens, the original primary replica remains down because the log reader does not move forward until all secondary databases are brought back online or until the failing secondary replicas are removed from the availability group. This ensures that on failover, all rows replicated to a subscriber also are present at the new primary. When a published database is a member of an availability group, by default the log reader will not process log records that have not already been hardened at all availability group secondary replicas.
The Log reader Agent has the following changes.
Some retry capabilities have been built in to the agents to handle the expected disconnect from a published database when it fails over to a new primary. Those errors with severity greater than or equal to 16 will cause the agents to terminate. When the switch is used, the redirected publisher is returned immediately to the agent and execution of the validation stored procedure is bypassed.įailures returned from the validation stored procedure are logged in the agent history logs. A new command-line switch, BypassPublisherValidation, has been added to the Log reader, Snapshot, and Merge agents. However, if agent startup occurs frequently the overhead associated with the validation stored procedure may be deemed too costly. The Log Reader, Snapshot, and Merge agents were modified to query the distribution database for the redirected publisher and to use the returned availability group listener name, if a redirected publisher was declared, to connect to the database publisher.īy default, when the agents query the distributor to determine whether the original publisher has been redirected, the suitability of the current target or redirection will be verified prior to returning the redirected host to the agent. Three replication agents were modified to support Always On availability groups. General changes to replication agents to support availability groups After failover, a tracer token cannot be entered by using the Replication Monitor, however a tracer token entered on the new publisher by using Transact-SQL, is visible in Replication Monitor. It may not be redirected to an explicit node.Īfter failover to a secondary replica, Replication Monitor is unable to adjust the name of the publishing instance of SQL Server and will continue to display replication information under the name of the original primary instance of SQL Server. If a published database is a member of an availability group and the publisher is redirected, it must be redirected to an availability group listener name associated with the availability group. Republishing is only supported when transactional replication is combined with Always On availability groups. In an availability group, a secondary database cannot be a publisher.
When the replication agent restarts after failover, the connection will automatically be redirected to the new primary. Established connections through the availability group listener name will fail on failover. These entries redirect the originally configured publisher/database pair, making use of an availability group listener name to connect to the publisher and publishing database. When a published database is aware of Always On availability groups, the distributor that provides agent access to the publishing database is configured with redirected_publishers entries. Overview of replication with availability groups Publisher Redirection Always On availability groups helps provide high availability and other database recovery capabilities. SQL Server Replication, change data capture (CDC), and change tracking (CT) are supported on Always On availability groups. Applies to: SQL Server (all supported versions)
0 kommentar(er)
