dbcc Options for High Availability Systems

dbcc Options for High Availability Systems

Sybase Failover includes dbcc ha_admin, which addresses second point of failures. Second point of failures a situations in which a cluster operation fails because of problems with the high availability subsystem. For example, if you issue sp_companion ‘prepare_failback’ and the secondary companion crashes. dbcc ha_admin provides a method of backing out of the cluster operation. After dbcc ha_admin is complete, you can re-issue the cluster operation.

dbcc ha_admin should only by a System Administrator who is familiar with the high availability subsystem. Issuing this command at the wrong time may only further trouble an already problematic situation.

Table B-4 includes information about the dbcc ha_admin options.

**Table B-4: dbcc ha_admin options**
Option Name	Function	Syntax and Comments
rollback_failback	Rolls back the effect of sp_companion... prepare_failback and returns the companion to the failover mode. This command works irrespective of the results of prepare_failback command.	dbcc ha_admin (" ", rollback_failback) Where “ “ is a required empty placeholder Can only be used in failback mode. Any failback threads waiting for the resume command are killed with this command. You may need to perform platform-specific steps to prepare you companions for rollback_failback option. See the configuration chapter for your platform for more information. This command is only issued from the secondary companion.
rollback_failover	Rolls back the effects of failover from the primary companion, and returns it to normal companion mode. rollback_failover does not affect the secondary companion.	dbcc ha_admin (" ", rollback_failover) Where “ “ is a required empty placeholder This command can only be used in failover mode. You may need to perform platform-specific steps to prepare you companions for rollback_failover option. See the configuration chapter for your platform for more information. rollback_failover has no effect on the companion server that failed. The companion server that takes over the failed companion’s work load resumes normal companion mode. This command is only issued from the secondary companion. This command works even when failover marked the databases “suspect”
drop_failedoverdb	Only used in failover mode. drop_failoeverdb drops the failed-over databases that could not be dropped with the drop database command. This command also cleans up the master_companion of all the metadata relating to the dropped database	dbcc ha_admin (" ", drop_failedoverdb, database_name) Where “ “ is a required empty placeholder, and database_name is the name of the database you are dropping. Use as a last resort, when you must drop a database to complete the load of another database.
clusterlock	Acquires or releases cluster-wide locks during a cluster operation.	dbcc ha_admin (" ", clusterlock, [acquire \| release]) For more information about cluster-wide locks and releasing them, see “Cluster Locks in a High Availability Node”.
state_machine	Moves the companion server to single-server mode.	dbcc ha_admin (' ', 'state_machine', 'halt') Where “ “ is a required empty placeholder. For information about using this option, see “Re-Installing installmaster and installhasvss”.
session	Invokes clients that are sleeping because of a failed sp_companion...resume. Clients that are invoked disconnect from the secondary companion and connect to the primary companion.	dbcc ha_admin (SYB_HACMP, session, "drop")

Table B-4: dbcc ha_admin options

Option Name

Function

Syntax and Comments

rollback_failback

Rolls back the effect of sp_companion... prepare_failback and returns the companion to the failover mode. This command works irrespective of the results of prepare_failback command.

dbcc ha_admin (" ", rollback_failback)

Where “ “ is a required empty placeholder

Can only be used in failback mode.
Any failback threads waiting for the resume command are killed with this command.
You may need to perform platform-specific steps to prepare you companions for rollback_failback option. See the configuration chapter for your platform for more information.
This command is only issued from the secondary companion.

rollback_failover

Rolls back the effects of failover from the primary companion, and returns it to normal companion mode. rollback_failover does not affect the secondary companion.

dbcc ha_admin (" ", rollback_failover)

Where “ “ is a required empty placeholder

This command can only be used in failover mode.
You may need to perform platform-specific steps to prepare you companions for rollback_failover option. See the configuration chapter for your platform for more information.
rollback_failover has no effect on the companion server that failed. The companion server that takes over the failed companion’s work load resumes normal companion mode.
This command is only issued from the secondary companion.
This command works even when failover marked the databases “suspect”

drop_failedoverdb

Only used in failover mode. drop_failoeverdb drops the failed-over databases that could not be dropped with the drop database command. This command also cleans up the master_companion of all the metadata relating to the dropped database

dbcc ha_admin (" ", drop_failedoverdb, database_name)

Where “ “ is a required empty placeholder, and database_name is the name of the database you are dropping.

Use as a last resort, when you must drop a database to complete the load of another database.

clusterlock

Acquires or releases cluster-wide locks during a cluster operation.

dbcc ha_admin (" ", clusterlock, [acquire | release])

For more information about cluster-wide locks and releasing them, see “Cluster Locks in a High Availability Node”.

state_machine

Moves the companion server to single-server mode.

dbcc ha_admin (' ', 'state_machine', 'halt')

Where “ “ is a required empty placeholder. For information about using this option, see “Re-Installing installmaster and installhasvss”.

session

Invokes clients that are sleeping because of a failed sp_companion...resume. Clients that are invoked disconnect from the secondary companion and connect to the primary companion.

dbcc ha_admin (SYB_HACMP, session, "drop")