Health monitor
The health monitor captures information about the database manager, database, table space, and table space containers.
The health monitor calculates health indicators based on data retrieved from database system monitor elements, the operating system, and database. The health monitor can only evaluate health indicators on a database and its objects when the database is active. You can keep the database active either by starting it with the ACTIVATE DATABASE command or by maintaining a permanent connection to the database.
The health monitor retains a maximum of 10 history records for each health indicator. This history is stored in the instance_ path\hmonCache directory and is removed when the health monitor is stopped. The health monitor automatically prunes obsolete history records when the maximum number of records is reached.
Health monitor data is accessible through health snapshots. Each health snapshot reports the status for each health indicator based on its most recent refresh interval. The snapshots are useful for detecting existing database health problems and predicting potential poor health of the database environment. You can capture a health snapshot from the CLP, by using APIs in a C or C++ application, or by using the graphical administration tools.
Health monitoring requires an instance attachment. If an attachment to an instance was not established using the ATTACH TO command, then a default instance attachment to the local instance is created.
In partitioned database environments, snapshots can be taken at any partition of the instance, or globally using a single instance connection. A global snapshot aggregates the data collected at each partition and returns a single set of values.
Usage notes
The health monitor is supported on all editions of the database.
On Windows, the service for the database instance needs to run under an account with SYSADM authority. You can use the -u option on the db2icrt command, or use the Services folder on Windows and edit the Log On properties to use an account with administrator privilege.
The health monitor process runs as a fenced mode process. These processes appear as DB2FMP on Windows. On other platforms, the health monitor process appears as DB2ACD.
The Administration server must be running on the system where the health monitor resides for notifications to be sent and alert actions to be run. If remote scripts, tasks, or contact lists are used, the Administration server on the remote system must also be started.
The tools catalog database is required only for creating tasks. If you do not use alert task actions for any health indicator, the tools catalog database is not required by the health monitor.