HP 3000 Manuals HP 3000 Manuals
Summaries
This section contains examples of the following summary report phases:
* Process log reason summary.
* SCAN START and STOP actual dates and times.
* Application overall summary.
* SCOPE coverage summary.
* Log file contents summary.
* Log file empty space summary.
Process Log Reason Summary
You must have PROCESS data in the log file.
This report helps you set the interesting process thresholds for SCOPE.
The report lists every reason a process might be considered interesting,
and thus get logged, along with the total number of processes logged that
satisfied each condition.
The following is an example of a process log reason summary report:
PROCESS SUMMARY REPORT 03/13/90 10:08 AM TO 03/30/90 8:48 AM
THERE WERE 382.6 HOURS OF PROCESS DATA
PROCESS RECORDS WERE LOGGED FOR THE FOLLOWING REASONS:
LOG REASON RECORDS PERCENT RECS/HR
--------------- ------- ------- -------
NEW PROCESSES 20276 14.6% 53.0
KILLED PROCESSES 19281 13.9% 50.4
CPU THRESHOLD 15338 11.1% 40.1
DISC THRESHOLD 1765 1.3% 4.6
TRANSACTIONS 226 0.2% 0.6
FIRST RESPONSE 10190 7.3% 26.6
RESPONSE TIME 6025 4.3% 15.7
WAIT ON CPU 3817 2.8% 10.0
WAIT ON DISC 1459 1.1% 3.8
WAIT ON MEMORY 1104 0.8% 2.9
WAIT ON IMPEDE 7455 5.4% 19.5
SHORT PROCESSES 27 0.0% 0.1
NOTE: A PROCESS MAY BE LOGGED FOR MORE THAN ONE REASON AT A TIME.
RECORD COUNTS AND PERCENTAGES DO NOT ADD UP TO 100 PERCENT
OF THE PROCESS RECORDS.
If DETAIL=ON, this report is generated each time a threshold value is
changed so you can evaluate the effects of that change. Each report
covers the period since the last report. A final report, generated when
the scan is finished, covers the time since the last report.
If DETAIL=OFF, then only one report is generated covering the entire
scanned period.
You can reduce the amount of process data logged by SCOPE by raising the
thresholds of the interest reasons that generate the most process log
records. To increase the amount of data logged, lower the threshold for
the area of interest.
In the previous example, you can decrease the amount of disc space used
for the process data (at the expense of having less information) by
raising the CPU threshold or setting the NONEW threshold.
If you want more information on processes that are getting poor response
time, you can lower the threshold for RESPONSE or FIRST.
The last category, SHORT PROCESSES, is slightly different from the
others. It lists the number of processes that were logged but that might
not have been logged if the NOSHORT threshold had been used.
SCAN Start and Stop
This summary report will be printed if any valid data was scanned. It
gives actual dates and times that the SCAN was started and stopped.
SCAN STARTED ON 01/03/90 12:40 PM
SCAN STOPPED ON 05/11/90 1:25 PM
Application Overall Summary
You must have APPLICATION data in the log file.
This report is an overall indicator of how much system activity is
accumulated in user-defined applications, rather than in OTHER
applications. If a significant amount of a critical resource is not
being captured by user applications, you might consider scanning the
process data for processes that can be included in user applications.
OVERALL, USER DEFINED APPLICATIONS ACCOUNT FOR
82534 OUT OF 112355 RECORDS ( 73.5%)
218.2 OUT OF 619.4 CPU HOURS ( 35.2%)
24.4 OUT OF 31.8 M DISC IOS ( 76.8%)
0.2 OUT OF 0.6 M TRANS ( 27.3%)
Performance Alarm Summary
You must use the PARMFILE command to specify process alarm definitions.
Setting ALARMS and DETAILS does not affect the generation of the
Performance Alarm Summary. See chapter 7, "Performance Alarms," for more
information.
Table C-2. Performance Alarm Summary
Alarm Count Minutes Description
1 1 25 CPU is monopolized by
sessions.
2 1 15 Batch is starved for
CPU.
3 0 0 Disc utilization.
4 9 80 Memory thrashing.
5 0 0 Runaway terminal.
Collector Coverage Summary
This report is printed if any valid global or application data was
scanned. It indicates how well SCOPE is being used to capture system
activity. If the percentage of time SCOPE was off is high, as in the
example below, then you should review your operational procedures for
starting and stopping SCOPE.
THE TOTAL TIME COVERED WAS 108/16:14:51 OUT OF 128/00:45:02
TIME LOST WHEN COLLECTOR WAS OFF 19/08:30:11 15.12%
THE SCOPE COLLECTOR WAS STARTED 45 TIMES
This report will be more compete if global detail data is included in the
scan. If only summary data is available, you can determine the time
SCOPE was stopped and started only to the nearest hour. (An appropriate
warning message is printed with the report if this is the case.)
The total time covered is determined by accumulating all the interval
times from the logged data. The OUT OF time metric is calculated by
subtracting the starting date and time from the ending date and time.
This should represent the total time that could have been logged. The
TIME LOST WHEN COLLECTOR WAS OFF metric is the total time less the
covered time.
The formats for the three times mentioned are as follows:
ddd/hh:mm:ss
where ddd are days and hh:mm:ss are hours, minutes, and seconds.
In the previous example, the total time was 108 days, 16 hours, 14
minutes, and 51 seconds.
The number of times SCOPE was started is equal to the number of times
SCOPE was restarted or the UTILITY command "SCOPE NEWPARM" was issued,
plus one.
Log File Contents Summary
This summary is printed if any valid data was scanned. It includes the
log file space and the dates covered.
------TOTAL------ --EACH FULL DAY-- -------DATES------- FULL
TYPE RECORDS MEGABYTES RECORDS MEGABYTES START FINISH DAYS
GLOBAL 29864 10.32 274.8 0.095 01/03/90 to 05/11/90 108.7
(NOTE RECS) 187 0.01 1.7 0.000
APPLICATION 113017 11.74 1040.0 0.108 01/03/90 to 05/11/90 108.7
PROCESS 138762 17.77 1300.1 0.166 01/22/90 to 05/11/90 106.7
DISC SPACE 214 0.06 1.0 0.000306 08/11/89 to 05/10/90 210.0
OVERHEAD 0.73
------- --------- ------- ---------
TOTAL 282044 40.63 2617.7 0.370
The columns can be explained as follows:
Column Explanation
TYPE The general type of data being logged. Two
special types exist:
* NOTES is the number of NOTES
generated (SCOPE notes plus USER
notes). These notes are actually
kept with the global data but are
listed separately in this report.
* OVERHEAD is the amount of disc space
occupied (or reserved) by the log
file versus the amount actually used
by the scanned data records.
If less than the entire log file was
scanned, OVERHEAD will include the
data records that were not scanned.
If the entire file was scanned,
OVERHEAD will account for any
inefficiencies in blocking the data
into the file plus any file-access
support structures.
It is normal for extracted log files
to have a higher overhead than raw
log files since they have additional
support structures for quicker
positioning.
TOTAL The total count and disc space scanned for
each type of data.
EACH FULL DAY The number of records and amount of disc
space used for each 24-hour period that
SCOPE runs.
DATES The first and last valid dates for the data
records of each data type scanned.
FULL DAYS The number of full (24-hour) days of data
scanned for this data type.
FULL DAYS may not be equal to the difference
between the start and stop dates if SCOPE
coverage did not equal 100 percent of the
scanned time.
The TOTAL line (at the bottom of the listed data) will give you an idea
of how much disc space you are using and how much data you can expect to
accumulate each day.
Log File Empty Space Summary
This summary is printed for each log file scanned.
THE GLOBAL FILE IS NOW 51.6% FULL WITH ROOM FOR 103 MORE FULL DAYS
THE APPLICATION FILE IS NOW 60.1% FULL WITH ROOM FOR 74 MORE FULL DAYS
THE PROCESS FILE IS NOW 89.1% FULL WITH ROOM FOR 13 MORE FULL DAYS
THE DISC SPACE FILE IS NOW 9.0% FULL WITH ROOM FOR 2432 MORE FULL DAYS
The amount of room available for more data is calculated based on the
amount of unused space in the file and the scanned value for the number
of megabytes of data being logged each 24-hour day (see "Log File
Contents Summary" ). If the megabytes-scanned-per-day values appear
unrealistically low, they are replaced with default values for this
calculation.
NOTE This report is made on a file-by-file basis. NOTES on the previous
report are included with the global data. OVERHEAD, which was
reported as combined overhead in the previous report, is calculated
for individual files in this report.
If you scan an extracted file, you will get a single report line since
all data types share the same extracted file.