Each DataCheck destination configuration connects to one source mirror member. Although the source member should not be changed, additional DataCheck configurations can be created to check against more than one source mirror member (or to check different sets of data from the same source).
Checking Data Between Failover Members
When checking between failover mirror members, the check is typically run with the backup failover member configured as the DataCheck destination for the following reasons:
-
The DataCheck destination uses more resources than the source in order to maintain the results of the check and other state information (which is itself journaled).
-
If the backup failover member is the DataCheck destination, the results are available for review on the backup if the primary failover member goes down.
Note:
In most configurations, it is assumed that the failover has already occurred and any review of the results probably happens after the failover decision point.
Whenever DataCheck loses its connection to the source, it retries the connection, waiting indefinitely for the source machine to become available again. If a mirror-based DataCheck is started on the destination when it was not the primary failover member, and that member becomes the primary, DataCheck stops rather than automatically try to reconnect. This prevents DataCheck from unintentionally running on the primary. For more information about reconnecting, see Starting/Stopping/Reconnecting DataCheck in this chapter.
Checking Data on Async Members
When mirror-based DataCheck is checking between a failover member and an async member, the async member is typically the destination. This is for the same reasons mentioned above (see Checking Data Between Failover Members) in regards to checking between failover members, but primarily because the results of the check should be stored on the async member during disaster recovery.
When there are two failover members, it is often desirable to create one DataCheck destination configuration on an async member for each of the two failover members as sources. The ^DATACHECK routine offers to create both for you, and offers settings for how they behave with respect to which of the two is the primary failover member.
Each DataCheck configuration has a setting to govern how it behaves based on the source failover member’s status as the primary member. The settings are:
-
No restriction
Checking both without restriction (the default) is desirable because it uses the async member as an agent to check both failover members without needing to run DataCheck between the failover members.
-
Check primary only (pause until DataCheck source is primary)
Checking against the primary only is desirable because the primary is the true source of the data for this async member.
-
Do not check primary (pause when DataCheck source is primary)
Checking against the backup is desirable because it does not consume resources on the production primary system.
For DataCheck configurations that are run manually (on demand) by a system administrator, these settings may not be of particular importance; they are more important for DataCheck configurations that are run continuously (or nearly so).
Any member may check another member without any particular relation. For example, if an async member is being used to check both failover members, it could also be used as the source of a check for other async members, thus avoiding the need to have any other async members check against the failover members.
