This section describes each mode in detail.
In this mode, Adaptive Server acts as a standalone server. A newly installed Adaptive Server is in single-server mode by default.
When both companions are running and are configured for failover, they operate in normal companion mode. This is the mode in which the day-to-day operations of Adaptive Server occur. For asymmetrical systems, this means that the primary companion can failover to the secondary companion. For a symmetric system, this means that either companion can fail over to the remaining companion.
Use suspended mode to temporarily suspend the companions from normal companion mode. In suspended mode, both servers act as single servers. Suspended mode is useful for performing system maintenance because you can start and stop the Adaptive Server and associated resources without risking failover.
Even though the companions cannot fail over, the nodes upon which they are working can still fail over; you must perform some platform-specific steps to suspend node failover. Also, before you shut down a companion in suspended mode, you must perform some platform-specific tasks. See the chapter for your platform for more information.
Many utilities and commands are severely restricted in suspended mode. See Appendix B, “Changes to Commands, System Procedures, System Databases, and New dbcc Commands, and Functions,” for more information.
Always suspend companion mode from the secondary companion
To suspend a companion from running in normal companion mode for any length of time (typically for maintenance), enter:
sp_companion 'secondary_server_name', 'suspend'
For example, to suspend primary companion MONEY1 from normal companion mode with its secondary companion PERSONEL1, issue the following from PERSONEL1:
sp_companion "MONEY1", suspend
The companion produces messages similar to the following:
Step: Server 'MONEY1' is alive and cluster aware Step: Access verified from Server:'MONEY1' to Server:'PERSONEL1' Step: Server 'PERSONEL1' is alive and cluster aware Step: Access verified from Server:'PERSONEL1' to Server:'MONEY1' Step: Companion servers configuration check succeeded Step: Access across the servers verified
Adaptive Server must enter the transitional failback mode to move from failover mode on the secondary companion to normal companion mode on the primary companion.
Failback mode is a planned even. That is, it is only done when the SA determines that the system is ready to resume normal operations. Use sp_companion “prepare_failback" to initiate failback and migrate the failed-over Adaptive Server to its original node. “Performing Failback” describes the steps necessary to perform failback.
To resume normal companion mode:
sp_companion "primary_server_name", resume
For example, to resume normal companion mode between primary companion MONEY1 and its secondary companion PERSONEL1, issue the following from MONEY1:
sp_companion "PERSONEL1", resume
The companion from which you issued the command produces messages similar to the following:
Step: Server 'MONEY1' is alive and cluster aware Step: Access verified from Server:'MONEY1' to Server:'PERSONEL1' Step: Server 'PERSONEL1' is alive and cluster aware Step: Access verified from Server:'PERSONEL1' to Server:'MONEY1' Step: Companion servers configuration check succeeded Step: Checkin to See if the remote server is up Step: Access across the servers verified Step: Synchronizing server logins from companion server Step: Synchronizing remoteserver from companion server Step: Synchronizing roles from companion server Step: Synchronizing server-wide privs from companion server Step: User information syncup succeeded
To permanently disable companion mode, enter:
sp_companion “server_name”, ‘drop’
The companion from which you issued the command produces messages similar to the following:
Step:Local server ‘MONEY1’ is alive and cluster aware Step:HA Versions and DLL check succeeded Step:Access across the servers verified Step: Removed the servers ‘MONEY1’ and ‘PERSONEL1’ for cluster config (return status = 0)
After completing this command the two Adaptive Servers are no longer companion servers and are running in single-server mode.
Drop is an irreversible operation. Once you have reverted the companion servers to single-server mode, you must dump, drop, and reload all user databases to reconfigure them as companions.
If the companion you drop is in a symmetric setup, the cluster automatically assumes an asymmetric setup between the companions.