dotnet
/
performance
зеркало из https://github.com/dotnet/performance.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность
performance/docs/scenarios-workflow.md

5.3 KiB
Исходник Постоянная ссылка Ответственный История

Scenario Tests Guide

Overview

Our existing scenario tests are under src\scenarios in this repo, where each subdirectory contains a test asset that can be combined with a specific set of commands to do measurements. Currently we have scenario tests for SDK, Crossgen, Blazor and other scenarios.

Running scenario tests

This is a general guideline on how the scenario tests are arranged in this repo. We will walk through it by measuring the startup time of an empty console template as a sample scenario. For other scenarios, refer to the following links:

Prerequisites

  • python3 or newer

    • some of the scenarios require requests python module to be installed. To install the required module run:

      python3 -m pip install requests

  • dotnet runtime 3.0 or newer

  • terminal/command prompt in Admin Mode (for collecting kernel traces)

  • clean state of the test machine (anti-virus scan is off and no other user program's running -- to minimize the influence of environment on the test)

Step 1 Initialize Environment

Before running the test, it is important to choose the right version of dotnet to test. Follow the guidance below to set up PYTHONPATH (to run our Python test harness) and dotnet directory for the desired test environment. This step is applicable to all scenarios and can only be run once for one environment.

Go to src\scenarios and run the following command:

Windows

Start a new PowerShell environment in Admin Mode and run:

cd src\scenarios
.\init.ps1

The next steps will need to run in the same Powershell environment. You can also specify custom dotnet directory or download a new dotnet to use. Add -Help option for more information.

Linux

Start a new bash terminal with Root Access and run:

cd src/scenarios
. ./init.sh

The next steps will need to run in the same bash environment. You can also specify custom dotnet directory or download a new dotnet to use. Add -h or -help option for more information.

Step 2 Run Precommand

Now you have PYTHONPATH set and dotnet to test in PATH, the next step is to run precommand to set up the specific test asset. Precommand is necessary for some scenarios and different test assets require different commands. NOTE: for each test asset, not all commands are supported. Please refer to Command Matrix for available scenarios.

For some scenarios (not all), pre.py runs a defined precommand before the test run, which can but not limited to set up the asset by either creating a new template or using a static template.

Format for running precommands:

cd <asset directory> # switch to the specific asset directory

Windows

py pre.py <command> <options>  # run precommand

Linux

python3 pre.py <command> <options>  # run precommand

In our startup time of an empty console template example, we can run

cd emptyconsoletemplate
python3 pre.py publish -f net9.0 -c Release

The above command creates a new dotnet console template in emptyconsoletemplate\app\ folder, builds the project targeting net9.0 in Release and publishs it to emptyconsoletemplate\pub\ folder.

Run python3 pre.py --help for more command options and their meanings.

Step 3 Run Test

Upon this step, the project source code should exist under app\ directory. There should be published output under pub\ if the precommand is "publish", and built output under bin\ if the precommand is "build". Now the test should be ready to run. NOTE: for each test asset, not all commands are supported. Please refer to Command Matrix for available scenarios.

test.py runs the test with a set of defined attributes. Format for running test commands:

Windows

py test.py <command> <test-specific options>

Linux

python3 test.py <command> <test-specific options>

In our startup time of an empty console template example, we can run

python3 test.py startup

The above command runs the published app under pub\ with specified iterations and measures its startup time.

Test report and traces are saved into emptyconsoleatemplate\traces\ directory.

Run python3 test.py --help for more command options and their meanings.

Step 4 Run Postcommand

post.py should be optionally executed to clean up the artifacts. It's the same command for all scenarios.

py -3 post.py

The above command removes app, bin, traces, pub, tmp directories if generated.

Command Matrix

Some command options are only applicable for certain test assets. Refer to the command matrix for each scenario category for a list of available command combinations: