The Pega Platform maintains statistics about the number of executions since startup. The system holds these statistics as property values in the Log-RuleUsage class, with additional detail in the Log-RuleUsage-Details class. During system shutdown, the system writes instances of these classes to the PegaRULES database, for later review and analysis.
Reports about these shutdown statistics can be useful in understanding which rules execute most often, and which execute rarely or not at all.
Note: Log-RuleUsage data, which covers only rules of those rule types that are converted to Java code by rules assembly, is different from Log-Usage data, which covers system and requestor performance.
The Pega Platform maintains statistics during operation, which are written to the PegaRULES database for later review and analysis during application server shutdown.
The rules assembly data presented by the Performance tool covers only your own requestor session. In contrast, these statistics cover all requestors of every type, with cumulative statistics since startup. This can be helpful in identifying the most frequently and least frequently executed rules (for those rule types that are assembled using rules assembly).
By "rule", this topic means "a Java class created by rules assembly". One Pega Platform activity named Alpha-Mort-.Recalculate might produce multiple rules, one for each variation in:
You can access these statistics with reports or database reporting tools, such as SQL Query Analyzer or Crystal Reports, operating with the pr4_log_rule_usage
table.
The value of the pzInsKey
key property of a Log-RuleUsage is concatenated from three other property values:
Log-RuleUsage
The Java class name identifies the rule type. Three examples are:
DECISIONTREE.RA_DECLARE_DECISIONTREE_RULE_
PARSE_DELIMITED_GETPROMPTKEY_362AD3D39240B644A4CFBE58EA5DA52E
RA_ACTIVITY_DATA_ADMIN_OPERATOR_ID_
SAVEDEVELOPERPREFERENCESXML_0C4D9291362765E0470388C76C1FD527
RA_HTML_FRAGMENT_SMARTPROMPT_
NOINPUT_1EF7F65A6B1B426D7A6E5A270B8DF4DD!
To select rows from a specific node, restrict reporting to those Log-RuleUsage instances that match a fixed pxSystemNode value.
Unlike most other concrete classes derived from the Log- base class, the pxCreateDateTime property is not part of the key of the Log-RuleUsage class.
The following properties are of primary interest in each snapshot row. All statistics are since startup of a node or since installation, covering all requestors of all types.
Property |
Description |
More |
pyLabel | Identifies the type of snapshot. The value "shutdown" is reserved. | |
pxClassName | Full Java class name for the rule-assembled rule. | rules assembly |
pxFamilyName | Rule type concatenated with visible key, converted to lowercase characters and converting special characters to underscore characters. For example:
activity_code_pega_list_inuserdesktop |
|
pxLastUse | Date and time that the rule was last executed. | |
pxNumPeeks | Number of times the class loader checked to see whether this entry was loaded. | |
pxNumInvalidate | Number of times the cache content was invalidated (unloaded), such that the next request forces rules assembly. | rules assembly |
pxNumLoads | Number of times this class was loaded. | |
pxNumReload | Number of times this class was reloaded without forcing a rules assembly. | |
pxRuleCount | Number of dependences that this rule has. | |
pxCreateDateTime | First date and time this rule was captured by this snapshot. | |
pxUpdateDateTIme | Most recent date and time this rule was captured by this snapshot. | |
pxSize | Size of the Java CLASS file in bytes. | |
pxSystemNode | Node ID (hash) on which this snapshot was captured. | Node ID |
pxUseCount | Number of times this rule was requested from the cache. |
No automatic processing purges rows of the pr4_log_usage
table. If your system has been shut down more than once, statistics for pxNumInvalidate, pxRuleCount, pxUseCount, pxNumReload, pxNumPeeks, pxNumLoads are cumulative since installation. Statistics for pxUpdateDateTime, pxLastUse are from the most recent shutdown.
Two prconfig.xml settings control this facility. The default values are:
<env name="usage/rules/shapshotonshutdown" value="true" />
<env name="usage/rules/committhreshold" value="100" />
To disable this facility, update the prconfig.xml file to set the value of the shapshotonshutdown
setting to false. To change the number of rows added during shutdown before a database commit operation, revise the value for the committhreshold
parameter.
Note: As an alternative to updating the prconfig.xml file, you can use Dynamic System Settings to configure your application. See Dynamic System Settings and the prconfix.xml file.
On a larger system with thousands of rules in use, set a larger committhreshold
value to reduce the shutdown delay.
Notes:
In the normal table mapping, the system saves these statistics a row of the pr4_log_rule_usage
table.
Using the System Management Application, you can review these statistics since startup while the system is up, rather than at shutdown, by clicking the applicable node.
If desired, you can save interim snapshots of most of these statistics as instances of the Log-RuleUsage class. Run the standard activity Log-RuleUsage.TakeSnapshot, and enter as parameter a snapshot identifier (other than the keywords "shutdown" and "recent", which are reserved). To simplify later retrieval and reporting, enter a snapshot identifier that indicates a date or date and time, such as DEC212006NOON. Unlike the shutdown snapshot, rows of this snapshot contain statistics accumulated since the most recent startup, that is, not consolidated with older snapshot values.