Stout is a simple tool to check performance stability of long running applications. E.g. it can be used to verify that a server application doesn't leak memory, or that it never exceeds specified CPU usage.
stout.exe -i <file path> [--version] [-h]
Parameters:
-i <file path>, --ini <file path>
(required) Location of ini file containing test configuration
--version
Displays version information and exits.
-h, --help
Displays usage information and exits.
Example:
# run test configured in d:\my_test.ini
stout.exe -i d:\my_test.ini
Tests are configured through simple ini file which specifies which processes are to be executed, which counters are to be monitored, and which limits are to be checked.
Stout starts the applications under test or attaches to running applications and then lets them run for some time to collect baseline data. This is controlled by DELAY and SAMPLING_TIME parameters.
Then it continues to monitor the running application and compares the collected metrics against the baseline data. In case that some limit has been violated it reports an error and optionally stops the test.
Here is an example of configuration file:
[STOUT::COMMON] ; Common parameters
DELAY = 5 ; Initial delay in seconds. When the tested apps are
; started, baseline assessment begins after the delay
; expires. If ommitted, default is 5000 (5 s)
SAMPLING_TIME = 60 ; How often will counters be averaged and evaluated, in
; seconds. This is also the duration of baseline assessment
; If ommitted, 60 s is used as default. Valid values are in
; range [10-3600]
ON_ERROR = LOG ; LOG | STOP - how to react if limits are violated. If LOG
; is specified, violation will be logged and test will
; continue. If STOP is specified, test will be aborted.
; LOG is the default value.
DURATION = 60 ; How long will test be executed, in minutes. When this
; period expires, test will be stopped. If set to 0, test
; will run until manually stopped. Default is 60 s
; metrics which are required for all tested apps are specified here
METRIC = MEM_WS.avg < 1% ; Working set for all apps must not increase more than 1%
METRIC = MEM_PB.avg < 1% ; Private bytes for all apps must not increase more than 1%
[STOUT::BACKENDS] ; collected data are written to specified backends
GRAPHITE=graphitesvr:9924 ; send to graphite
CSV="path\file.csv" ; write to CSV file
; now we configure tested apps
[consumer] ; each app has a symbolic name - consumer, in this case
START = consumer.exe ; start consumer.exe
COUNT = 3 ; run 3 instances. if ommitted, default is 1
METRIC = MEM_WS.avg < 1% ; working set must be constant
[producer] ; run the producer
ATTACH = producer.exe ; don't start it - attach to an existing instance
; In this case, COUNT is attached.
; if multiple instances are running, process id can be
; specified, e.g. 'producer.exe:5624'
METRIC = CPU.avg < 10 ; avg CPU usage must not exceed 10%
METRIC = CPU.max < 50 ; peak CPU usage must not exceed 50%
METRIC = MEM_WS.avg < 1% ; working set must be constant
Following table lists all the metrics that are collected.
| Name | Description |
|---|---|
MEM_WS |
Working set size. If using absolute limit, it is specified in kB |
MEM_PB |
Private bytes. If using absolute limit, it is specified in kB |
CPU |
CPU usage. if using absolute limit, it is specified in % |
Checks can be done against average, min and max value and standard deviation:
METRIC = CPU.avg < 10 ; avg CPU usage must not exceed 10%
METRIC = CPU.max < 50 ; peak CPU usage must not exceed 50%
; avg / min / max / stddev are supported
Collected data can be logged. Different backends can be used (file, console...)
This is specified in STOUT::BACKENDS section of configuration file.