HP 3000 Manuals HP 3000 Manuals
Run Time Monitor Operation
RTM is a tool designed to aid in migrating applications from an MPE
V/E-based system to an MPE/iX-based system. The task of RTM is to
monitor executing applications for occurrences of predefined events. An
event is defined as an MPE V/E intrinsic feature that will be
incompatible when executed on an MPE/iX-based system.
RTM is made up of three parts:
* A controlling program, RTMSYS.PUB.SYS, used to enable RTM and to
specify the event classes to monitor.
* A report program, RTMREP.PUB.SYS, used to generate reports for
monitored applications.
* An SL, RTMSL.PUB.SYS, used by RTM to control event monitoring.
RTM uses the MPE V/E System Logging facility to log events during program
execution. Therefore, system logging must be enabled for RTM to log
events. To determine if RTM is operating, execute an event that you know
will be logged, close the current log file and create a new log file with
:SWITCHLOG, and generate a report with RTMREP.
RTMSYS
A user with SM or OP capability can use RTMSYS to enable or disable RTM
and monitoring of specific event classes. These classes are described in
Table C-1 .
Table C-1. Run Time Monitor Event Classes
-----------------------------------------------------------------------------------------------
| | | |
| Class Number | Class Name | Description |
| | | |
-----------------------------------------------------------------------------------------------
| | | |
| 0 | Logging Enabled | Enable/disable RTM operation |
| | | |
| 1 | NM Events | Enable disable logging of uniquely NM events |
| | | (in addition to events common to both CM and |
| | | NM. |
| | | |
| 2 | Command Intrinsic Events | Enable/disable logging of any incompatible |
| | | command executed by the COMMAND intrinsic. |
| | | Commands are not associated with a specific |
| | | mode. Therefore, the NM Events (class 1) does |
| | | not effect this event class. |
| | | |
| 3 | FFILEINFO Intrinsic | Enable/disable logging of FFILEINFO related |
| | Events | incompatibilities. There are several |
| | | differences relevant to FFILEINFO mainly |
| | | associated with device information. |
| | | |
| 4 | FGETINFO Intrinsic | Enable/disable logging of FGETINFO related |
| | Events | incompatibilities. |
| | | |
| 5 | FCONTROL Intrinsic | Enable/disable logging of FCONTROL related |
| | Events | incompatibilities. |
| | | |
| 6 | General File System | Enable/disable logging of all other file |
| | Events | system incompatibilities. |
| | | |
| 7 | | Reserved for future use. |
| | | |
| 8 | Miscellaneous Events | Enable/disable logging of events that do not |
| | | fall under the other event classes. |
| | | |
-----------------------------------------------------------------------------------------------
RTMREP
A report of events can be generated by using the RTMREP program. RTMREP
reads the specified system log files and generates a report to the formal
file designator RTMLIST of the events found. Two report formats are
available:
* A brief report (the default) is a program summary which shows the
events generated by each program. A counter shows the number of
times that a particular call caused the event. The report is
sorted by date, program name, and event class.
* A detailed report shows all events logged, sorted by date, program
name, and event class. All information in the log record is
formatted and displayed in chronological order.
Appendix D "Incompatibilities" contains detailed descriptions of all
events detected by RTM and reported in RTMREP. The descriptions include
possible causes of the event, as well as recommended actions you can take
to work around the detected incompatibility.
RTMSL
RTM uses an SL file, RTMSL.PUB.SYS, to find areas of incompatibility by
intercepting calls to specific system procedures. The SL is bound to the
program at load time and allows you to control which events are logged.
The relevant information is logged by RTM, which then completes the
procedure call by calling the actual procedure in SL.PUB.SYS. With RTM
enabled, the LOADER inserts RTMSL.PUB.SYS into the load sequence when any
program not residing in PUB.SYS is executed or allocated.
Managing Run Time Monitor Stack Space Usage
When RTM is enabled, your programs will use slightly more stack space.
The amount of additional stack depends on the intrinsics called by your
program. Intrinsics with more parameters will incur more stack overhead
when RTM is enabled. The additional stack overhead ranges from 8 to 60
words. Your applications will use more stack from the time the intrinsic
is called until control is returned to your program. In other words,
your application does not pay a penalty in stack overhead during the
entire time it is executing, and the stack overhead is not cumulative for
all the intrinsics your application calls. Instead, the additional
overhead impacts your application only when RTM intercepts the call to a
potentially incompatible procedure. Nevertheless, it is possible that
the additional stack space used by RTM will cause your program to abort
with a stack overflow message. If this happens you have 2 options.
1. You can attempt to run your program using the ;NOCB option on the
:RUN command. You will also want to make sure you have specified
the maximum value for the ;MAXDATA= option on the run command or
when the program was prepped with the :PREP command.
If this option does not work or is too cumbersome, because you
would have to modify several system UDCs to make it work, attempt
to use option 2.
2. You can still use RTM to monitor other programs on your system
while continuing to use the application which is aborting. To do
this you, should disable RTM and make sure the program in question
is not being run and is not allocated. Having accomplished this
you may allocate the program using the MPE :ALLOCATE command and
enable RTM. Programs which are allocated before RTM is enabled
will not be loaded through RTMSL.PUB.SYS and therefore, will not
incur any overhead due to RTM. Unfortunately, RTM will not be able
to monitor these programs so you will need to rely on OCA for your
migration information.
Managing Run Time Monitor Disk Space Usage
RTM uses the MPE V/E System Logging facility to keep track of migration
events as they occur on your system. As log file records are generated,
RTM will consume disk space on your system. The rate of consumption
varies based on several different factors. These factors include the
following:
* Other activities also being logged on your system (for example,
console logging).
* System load (for example, numbers of users, extent of activity on
your system).
* Which RTM event classes have been enabled using RMTSYS.PUB.SYS.
* Period of operation of RTM.
* Number of migration issues encountered in the programs being
monitored.
If you are concerned about disk space usage on your system you may wish
to monitor the system log files when you first enable RTM (or when you
change the event classes being monitored via RTMSYS.PUB.SYS). One easy
way to do this is to issue the following :LISTF command.
______________________________________________________________________________________
| |
| |
| :LISTF LOG####,PUB,SYS,2 |
| ACCOUNT= SYS GROUP= PUB |
| |
| FILENAME CODE LOGICAL RECORD SPACE |
| SIZE TYP EOF LIMIT R/B SECTORS #X MX |
| |
| LOG1327 1022W VB 27165 1023 1 8192 16 16 |
| LOG1328 1022W VB 13012 1023 1 4096 8 16 |
| LOG1329 1022W VB 5402 1023 1 2048 4 16 |
| |
______________________________________________________________________________________
By totaling the SECTORS column you can determine how much disk space is
being consumed by the system log files (in the example above, 14,336
sectors have been used).
If you find that disk space is being consumed too rapidly the following
corrective steps are recommended:
1. Store system log files on tape for future analysis as they are
generated.
2. Disable monitoring for event classes once you feel you have
sufficient data on events in those classes.
3. Temporarily disable other types of system logging while using RTM.