DB2 10.5 for Linux, UNIX, and Windows

HADR log spooling is now enabled by default

HADR log spooling is now enabled by default. The hadr_spool_limit database configuration parameter, which specifies log spooling behavior, now has a default setting of AUTOMATIC.

Details

Log spooling is enabled for any database that you create in Version 10.5. For existing databases, what happens during an upgrade to Version 10.5 depends on whether or not you are using the DB2® pureScale® Feature:

For DB2 pureScale databases, the hadr_spool_limit parameter is set to AUTOMATIC. Because HADR was previously not supported with the DB2 pureScale Feature, this change has no impact.
For all other DB2 databases, the value of the hadr_spool_limit parameter is not changed.

Introduced in Version 10.1, HADR log spooling allows transactions on the primary to make progress without having to wait for the log replay on the standby. Testing has shown that spooling can significantly reduce any impact that HADR might have on the primary's workload but has a negligible impact on a standby's replay performance.

Resolution

If you do not want HADR log spooling to be operational, set the hadr_spool_limit database configuration parameter to 0. This parameter takes effect only on the standby. You need to deactivate and reactivate the standby to make it pick up the new value. For the primary, stop HADR, then start HADR to pick up the new value (while keeping the database online) so that when the database becomes a standby, it uses the new value.

To ensure that HADR log spooling can work properly, ensure that you provide adequate disk space for the active log path of the standby database.