Failover for High-availability clusters in an Enterprise Replication environment

About this task

If you configure connection management for failover, Connection Managers can promote a secondary server to the primary-server if the primary server fails. If connection management is not configured to control failover, the onmode -d make primary command can promote a secondary server to the primary-server role. In either of these cases, Enterprise Replication automatically connects to the new primary server.

If the primary server fails, and you manually change a secondary server to a standard server, you must complete the following steps to prevent Enterprise Replication from starting on all servers in cluster.

Run the following commands on the secondary server:
  1. onmode -s
  2. onmode -d standard
  3. cdr start

If Enterprise Replication is running on the secondary server, and you want to restart the server that was the primary server, but without Enterprise Replication and high-availability cluster replication, run the oninit -D command. You can then stop Enterprise Replication on the standard server and reestablish the primary server.

First, run the following commands on the standard server:
  1. cdr stop
  2. onmode -d secondary primary_ha_alias
Second, run the following commands on the primary server:
  1. oninit
  2. cdr start

To split an active cluster into two standalone servers, you must restart the database servers with the oninit -D command to prevent Enterprise Replication from starting on either server after they are split.

To remove a server from a cluster, run the cdr delete server force ha_alias command, where ha_alias is an Enterprise Replication group name, to remove Enterprise Replication from that server. For example, the two HDR servers are being split and the secondary server is to be used for reporting purposes. After the report processing is complete, HDR can be reestablished. cdr delete server shows how to remove a secondary server from a high-availability cluster and Enterprise Replication.
Table 1. Removing the Secondary Server from a cluster and ER
Step On the Primary On the Secondary
1. onmode -d standard secondary_ha_alias
2. Run onmode -d standard
3. Run cdr delete server -f ha_alias
If the HDR primary server has problems communicating to its secondary server, Enterprise Replication is in a suspended state until one of the following actions is taken:
  • Resolve the connection problem between HDR pairs.
  • Convert the primary server to standard mode.