NTS system parameters are used to do the following:
In addition to enabling and disabling certain NTS functions by setting system parameters, you can set or modify certain system values by using the SYSPARMS command. This lets you improve or modify NTS operations to suit your installation requirements. In most cases, the default values supplied by NTS should be adequate.
The NTS functions or parameter settings that you can configure and the associated SYSPARMS operands are listed in the following table.
|
Configurable Function or Setting |
Related Operand |
|---|---|
|
Collection of accounting data |
NTSACCT |
|
Collection of resource statistics |
NTSRSTAT |
|
Logging of active sessions at shutdown |
NTSCLOSE |
|
Intensive message logging |
NTSINTSV |
|
Notification of MAI sessions |
NTSMAISV |
|
Generation of NTS events |
NTSEVENT |
|
Queuing of NTS CNM requests |
NTSCNMQ |
|
Consolidation of trace final queue buffers when first wrap occurs |
NTSTRBFX |
|
Presentation of MAI sessions to the NTS user exit |
NTSMAIEX |
|
Correlation of data |
NTSCINTV |
|
Trace activity |
NTSMAXTR |
|
Session keep counts |
NTSSKEEP |
|
VTAM session and trace data buffers |
NTSSAWBF |
|
Resource statistics collection intervals |
NTSRSINT |
|
GMT/Local Timestamp in SMF T39 Record |
NTSSMFTM |
For more information about the SYSPARMS command syntax, and the valid operand values and their significance, see the Reference Guide.
To enable the collection of NTS session accounting data, use the NTSACCT operand. Collection of this data can be selective (the default) or global.
When session awareness processing is active, you can specify NTSACCT=NO only.
To change to any other value
NTSACCT=SELECTIVE is set by default. This means that accounting data for a session is collected only if the DEFCLASS ACCT operand is set to YES in the SAW class definition associated with the session.
To collect NTS accounting data for a particular session class
If you enable or disable the accounting function globally, SAW class definition accounting options are ignored.
To enable or disable the accounting function globally, specify NTSACCT=ALL or NTSACCT=NO.
To enable the collection of NTS resource accounting and RTM statistics, use the NTSRSTAT operand. This operand globally enables or disables resource statistics collection when you specify a value of YES or NO. The default is NO.
Also consider the following actions:
To change from NTSRSTAT=NO to NTSRSTAT=YES
Important! Carefully evaluate the requirements of your installation for resource statistics collection, because summarizing many resources may not give useful results and may adversely impact the performance of NTS.
The rules governing statistics collection are as follows:
Where a resource class specifies that statistics are collected, NTS accumulates resource statistics for resources that match the class, at the level of the resource class. For example:
If you have enabled the collection of statistics for a particular resource, NTS automatically uses SAW data to monitor the availability of that resource. A resource is considered available if it is participating in a session with the SSCP of the domain in which it is defined.
If NTS is monitoring resource availability, it automatically passes SMF records that indicate changes in the status of a resource to the NTS User Exit, if you have defined one.
When your region is being shut down, all NTS activities must cease. It is likely that a number of sessions will be active and that session data collected by NTS for those sessions will not be logged.
To treat these residual sessions as ended for the purpose of logging, set the NTSCLOSE operand of the SYSPARMS command to YES. The sessions are queued for output processing and the NTS class definitions checked to determine whether logging is required.
Because there is only a small delay (approximately 10 seconds) between the time that NTS is notified of the impending shutdown and the actual termination of your region, this setting is useful only when the residual session count is small.
An alternative method of closing sessions is available through the SAWARE STOP CLOSE command.
Note: To improve performance during NTS logging, operate the NTS database using VSAM Local Shared Resources (LSR) and deferred I/O capabilities.
NTS receives large quantities of data from VTAM and may not process data that it cannot understand. For example, NTS cannot collect trace data indefinitely for a session of which it has no knowledge. As a result, at some stage it purges such data. At other times, NTS may receive data that is not in the expected format and this data is discarded. During normal operation, these kinds of data are purged on a regular basis and may go unreported by NTS.
If you suspect that data is missing, to aid problem detection, you can enable intensive message recording to see if NTS is discarding any data.
To enable intensive message recording, set the NTSINTSV operand to YES. This creates log messages whenever the conditions of data inconsistencies arise.
To enable NTS to be aware of MAI sessions, specify NTSMAISV=YES (the default is NO).
For trace and accounting data to be available for an MAI session, collection must be requested for the primary half-session component of an MAI virtual session. When trace data is received for the primary half session and you have requested trace or accounting data collection for MAI sessions, NTS indicates that the MAI session has such data available. If you request the display of trace or accounting data for an MAI session, primary or secondary, then the data collected for the primary half session appears.
If you specify NTSMAISV=NO when the interface is already active, NTS retains knowledge of existing MAI sessions, but is not notified of any new MAI sessions.
To enable NTS event generation
After NTS has built a session record for an active session, the future processing for that session is fixed by the various values extracted from the matching class or classes. However, you may need to modify such processing options under certain circumstances, especially since sessions can remain active for extended periods.
To modify the processing options for an active session, use the NTSMOD command.
The NTSMOD command lets you do the following:
Example: NTSMOD Command
This is an example of the use of the NTSMOD command.
NTSMOD NAME=CICS TRACE=(4,50) LOG=FORCE
In this case, the following occurs:
When you review CICS session data at a later stage, the display of an F next to the end time for each session on the NTS Session List Panel indicates that these sessions were force-logged before they ended.
For more information about the NTSMOD command and its operands, see the online help.
Note: If you issue the NTSMOD command with neither the TRACE nor the LOG operand specified, then the sessions specified by the NAME operand are listed, so that you can determine the scope of the command prior to making any modifications.
One of the primary functions of NTS is to gather data from a number of sources and correlate it at session level.
The sequence in which data arrives and the interval between such arrivals, is beyond the control of NTS. Under certain circumstances, such as a network failure, anticipated data may not arrive.
To protect NTS from waiting indefinitely for such session data, there is an interval defined that represents the time limit for data correlation. The default correlation interval is 30 seconds.
To change the default correlation interval, use the NTSCINTV operand.
We recommend that the length of the correlation interval be kept constant throughout the network.
The STRACE command is used to start and stop global or specific NTS tracing. This command provides operands that let you select the precise session trace activity that you want. For more information about these operands, see the online help.
Global tracing consumes large amounts of system resources. To avoid this, NTS provides parameters to limit the number of outstanding trace requests.
The PIU operand of the STRACE command lets you do a PARTIAL or FULL tracing. In most instances, PARTIAL tracing provides sufficient data for problem determination.
Important! Before you issue a request to trace a complete RU from VTAM, take note that an RU can be very large.
To set the maximum number of specific trace requests that can be outstanding at any time, specify the limit in the value of the SYSPARMS NTSMAXTR operand.
NTS rejects any attempt to issue a specific trace request that would result in the value set for NTSMAXTR being exceeded.
To impose limits on the number of specific trace requests that can be outstanding, use the NTSMAXTR operand.
To set the trace queue depths for the initial and final trace queues for a session, set the values in your SAW class definitions. These values determine the maximum number of PIUs that can be stored for a session at any given time.
To override the values in the SAW definition, issue the SYSPARMS NTSMAXTP command.
Note: After you define values for NTSMAXTP, you cannot define a new SAW class with trace queue depths exceeding these values or change the trace queue depths of an existing SAW class to be greater than the values set for this system parameter.
The session keep count refers to the number of session incidences that are stored concurrently in the NTS database for any session name pair. The default session keep count is 10.
To modify the default count, use the NTSSKEEP operand.
Note: The session keep count is used the first time a session incidence for a new name pair is written to the database only. The value is subsequently stored with the records in the database. To modify this value, use the NTSDBMOD command.
When NTS session awareness processing begins, requests are sent to VTAM specifying the number and size of the buffers to allocate for the collection of session awareness and session trace data.
To modify these values, use the NTSSAWBF and NTSTRCBF operands.
NTS allocates the following by default, which should be adequate for normal usage:
However, during times of exceptionally heavy trace activity, the allocation may be insufficient.
When These Allocations are Insufficient
If NTS cannot service the data buffers quickly enough, VTAM overwrites the data in the oldest, unprocessed buffer, with the result that you lose data. NTS can detect this data loss and notify operators by issuing a monitor message.
In times of intense system activity, you may lose some trace data in this way.
Overcoming the Problem
If buffer overrun conditions occur, allocate a larger number of smaller buffers, rather than a smaller number of larger buffers. If more buffers are available to VTAM, they are likely to be available to NTS at any time.
Other factors influence the delivery of data to NTS, especially the ability of the operating system to dispatch data. If its dispatching priority is too low, it may not be able to service large amounts of trace data in times of intensive activity. You need to check the dispatching priority and ensure that it is set just below that of VTAM.
Resource statistics are collected and presented by NTS as counts of events that occurred in a specified time interval. Statistics gathered during different intervals can be compared for the purpose of network performance monitoring and analysis.
The valid range for the resource collection interval is 1 to 480 minutes (8 hours), and the default is 30 minutes.
To configure the duration of the interval, use the NTSRSINT operand to set the value you require.
To set the value of the number of intervals that can occur before NTS overwrites the statistics collected for the oldest interval, use the DEFCLASS RESOURCE LIMIT operand, or set the global default by using the SYSPARM NTSRSLIM operand. The valid range of values for this operand 0 to 255; the default is 16.
Important! High settings for this operand can consume large amounts of storage.
All application timestamps records in type 39 SMF records consist of the first four bytes of the system TOD clock value, plus a 4-byte signed number for the time zone adjustment value, in seconds. By definition, the first four bytes represent GMT time in approximately 1-second intervals. However, Tivoli NetView (NLDM) writes the first four bytes of these timestamps in local time.
The default is GMT.
To maintain the NTS database, use the NTSDBMOD command. This command lets you do the following:
Example: Maintain the NTS Database
This is an example of the use of the NTSDBMOD command:
NTSDBMOD PRINAME=CICS KEEPDATE=01/03/31
In this case, all stored records for sessions with the primary name of CICS that predate April 1, 2001, are deleted.
For more information about the NTSDBMOD command and its operands, see the online help.
After the first session incidence has been recorded for a session name pair, the master record contains the session keep count for that name pair.
To display this value, use the SHOW SKEEP command.
To modify this value, use the NTSDBMOD command.
To delete all records for the session name pair, including master and cross-reference records, set a new session keep count of zero in the NTSDBMOD command.
The NTSDBMOD command lets a generic name specification permit mass update and deletion with a single command.
You can convert NTS session records to SMF type 39 format to use the output in report-generating applications. To do this, you must specify the name of an NTS user exit in the SAW parameter group.
NTS automatically passes SMF-formatted records to this exit. The supplied exit can be customized to perform further processing of the data before the data is passed to SMF and finally to a report-generating application, such as SAS, to produce statistical reports based on the raw NTS data.