Verify that InfoSphere® CDC system
parameters are set to appropriate values so that query plans can use
shadow tables
Procedure
To ensure that the required InfoSphere CDC system
parameters are set correctly:
- Log in to the cdc-server-host as a cdc-user.
- Change to the bin folder for InfoSphere CDC for DB2® for
LUW by issuing the following command:
$ cd /cdc-installation-dir/bin
- View the current settings for the system parameters that
are not set to their default values by issuing the dmset command
as follows:
./dmset -I cdc-instance-name
- Check whether the value of the acceptable_latency_in_seconds_for_column_organized_tables system
parameter was modified by issuing the dmset command
as follows:
./dmset -I cdc-instance-name
acceptable_latency_in_seconds_for_column_organized_tables
The
default setting for the acceptable_latency_in_seconds_for_column_organized_tables system
parameter is 5 seconds. The acceptable_latency_in_seconds_for_column_organized_tables system
parameter controls the interval between batched commits when InfoSphere CDC replication
is active. The value of the acceptable_latency_in_seconds_for_column_organized_tables system
parameter must be less than the value of the CURRENT REFRESH AGE special
register. If the interval between batched commits is similar to or
larger than the value of the CURRENT REFRESH AGE special register,
latency might regularly become out-of-date for short periods. When
latency becomes out-of-date, queries are routed to base tables intermittently.
- Ensure that the maintain_replication_mqt_latency_table system
parameter is set by issuing the dmset command
as follows:
./dmset -I maintain_replication_mqt_latency_table
For InfoSphere CDC to
update the SYSTOOLS.REPL_MQT_LATENCY table, you must set the maintain_replication_mqt_latency_table system
parameter. By default, this parameter is set to false,
and InfoSphere CDC does
not send latency information to DB2.
In the absence of latency information that is required to make routing
decisions, shadow tables are not used.
- Ensure that the mirror_auto_restart_interval_minutes system
parameter is set to a value greater than 0 by issuing
the dmset command as follows:
./dmset -I mirror_auto_restart_interval_minutes
The mirror_auto_restart_interval_minutes system
parameter determines the interval in minutes between attempts to restart
a persistent subscription when replication for that subscription fails
due to a recoverable error. A value of 1 or 2 seconds
is suitable for most environments.