Analyst Overview

When an Analyst is started, the following processes take place:

  • The Process Manager (IRPROMGR) receives the START message specifying the Analyst rule and then runs an IRAUTOAN process, passing it the information.

  • The Process Manager sets the Analyst processes initial priority for the IRAUTOAN object from the PROGNOSIS Configuration SET PRIORITY parameter.

  • Parsing the Rules - Then the Analyst process starts, it parses the specified rule file.

  • Problem Output Files - Each Analyst started will create and own a set of output files that hold the contents of the problem record.

Parsing the Rules

The Analyst process will parse the rule file specified in the startup command for syntax and then for configuration. If a syntax error is encountered in the rule file, a message is returned to that effect to the user interface session that started the Analyst. At this point, the Analyst is unable to continue and the process will terminate itself.

Once the file is parsed successfully, the Analyst process will build a configuration request for the Process Manager based on any RECORD, ACTION or IF statements present. These statements set out the conditions under which the Analyst should begin problem actions and are based on values of Records and fields.

The following lines are an Analyst extract that show a simple rule:

SECTION RULE_DEF
RULE CpuBusy PRIMARY
RECORD NTCPU
WHERE NTCPU.BUSY > 80
REFRESH 60 SECONDS

ACTION
WHERE ALL
LOG 0001 IMMEDIATELY
END_ACTION
END_RULE
END_SECTION

In the above example, the Analyst will send a request to the Process Manager for the NTCPU record, but only where the CPU Busy value is greater than 80%. Anytime that the Process Manager finds an NTCPU record that fits these criteria it will pass it to the Analyst which will begin processing the rule from the ACTION statement.

Note that in this example the action statement itself contains another Where Clause. A Where Clause can be placed anywhere inside a rule and serves to further filter the data being collected via the RECORD statement.

The REFRESH parameter is included in the configuration to inform the Process Manager of how often it should refresh the data.

Only one RECORD statement can exist in a rule.

Problem Output Files

Each Analyst started will create and own four output files that hold the contents of the various Problem records, tracking the progress of the Analyst’s problem resolution. These files are named as F0010<nnn>, with <nnn> being an incrementing 3-digit number. Each Analyst reserves five files (one is unused) for itself. So, the first Analyst started will use 001, 002, 003 and 004. The second Analyst started will use 006, 007, 008 and 009 and so on.

The first file, F0010001 holds PrognosisAutomationProblem (PROBLEM) records of the type PRI_RULE. This type of problem record is generated for every Primary rule contained in the Analyst file. This file primarily used to monitor the state of all Primary rules. A Display using the PROBLEM record with a Where Clause of REC_TYPE = PRI_RULE can be used to see all active (but not necessarily enabled) Primary rules.

The second file, F0010002 holds all PrognosisAutomationProblem (PROBLEM) records with a type of DETAIL. These problem records are the result of Analyst detected problems and are written to this file as a result of any LOG commands executing from inside the rules file. A LOG command is a simple mechanism by which TEXT can be output from within a rule.

The third file, F0010003 holds all PrognosisAutomationProblemSummaryOpen (PROBSUM) records. These records contain a summary of each of the open problems on the system at any point in time. They are created as the result of LOG PROBLEM_SUMMARY commands executing from inside the rules file.

The fourth file, F0010004 holds all PrognosisAutomationProblemSummaryClosed (PROBSUMC) records. These records contain a summary of each of the last <n> closed problems. They are created when problems stored in the F0010003 are closed. When this occurs, the problem is deleted from the F0010003 file and added to the F0010004 files.

Setting the Output File Size

The size of the output files for the PrognosisAutomationProblem (PROBLEM), PrognosisAutomationProblemSummaryOpen (PROBSUM) and PrognosisAutomationProblemSummaryClosed (PROBSUMC) records can be set through the 'irautoan' ini file. After this setting is made, the output file will wrap when the specified size limit is reached and new problem data will then start to overwrite older data. Unfortunately, a recommended size cannot be given as the rate of growth is solely determined by the amount of output generated for a specific system, which in turn will be affected by items such as the number and complexity of the Analyst rules being used.

To set the file size go to the [THRESHOLD] section of the 'irautoan' ini file (despite the [THRESHOLD] title, this section applies to both Analysts and Thresholds). The 'irautoan' file is named IRAUTOAI on HPE NonStop and irautoan.ini on all other platforms (located in the Prognosis/Server/Configuration folder path). Under the [THRESHOLDS] section the following statements can be adjusted as required:

[THRESHOLD] Statement

Description

OpenSummaryFileMaxSize = <sizemb>

Sets output file size for the PrognosisAutomationProblemSummaryOpen (PROBSUM) record. Enter as a decimal value (e.g. 0.01 for 10KB).

If no value is added in this statement then a default of 0.5MB will apply

ClosedSummaryFileMaxSize = <sizemb>

Sets output file size for the PrognosisAutomationProblemSummaryClosed (PROBSUMC) record. Enter as a decimal value (e.g. 0.01 for 10KB).

If no value is added in this statement then a default of 0.5MB will apply

ProblemFileMaxSize = <sizemb>

Sets output file size for the PrognosisAutomationProblem (PROBLEM) record. Enter as a decimal value (e.g. 0.01 for 10KB)

The value added will apply to all Thresholds and Analysts. If no value is added in this statement then a default of 1MB will apply. For individual Analysts (not Thresholds) either the set or default value can be overridden by using the MAX_SIZE statement in the CONFIGURATION ( Analyst Rule Configuration and Global Variables) section of the rules file.

ConditionFileMaxSize = <sizemb>

Sets the size of file used to store condition status (PROBLEM.RECTYPE = “PRI_RULE”). Enter as a decimal value (e.g. 0.01 for 10KB).

If no value is added in this statement then a default of 0.5MB will apply

MaxDays = <days-of-data-to-keep>

Specifies the number of days of data to keep.

The value added will apply to all Thresholds and Analysts. If no value is added in this statement then a default of 0 days will apply. For individual Analysts (not Thresholds) either the set or default value can be overridden by using the MAX_DAYS statement in the CONFIGURATION ( Analyst Rule Configuration and Global Variables) section of the rules file.

Example:

[THRESHOLD]
OpenSummaryFileMaxSize = 0.5
ClosedSummaryFileMaxSize = 0.5
ProblemFileMaxSize = 1.0
ConditionFileMaxSize = 0.5
MaxDays = 7.0

The Command Server 'F' Files

There are two other 'F' files that are related to the Analyst, F0000001 and F0000002. There will only ever be one set of these files no matter how many Analyst processes are started. These files hold the output from the command server process that is performing the actual automation.

The data in F0000001 is available for display through the AUTOOUT record and contains the output of every Command run, including those run via EXEC statements.

The data in the F0000002 is available for display through the CMDLOG record and contains a summary of all the Commands that have been run, along with the user who initiated them and the result (success or failure) of the command.

Once all the 'F' files have been created and opened, the Analyst starts to parse (read in) the rules.

Provide feedback on this article