Skip to content

timothyb89/stackviz

Repository files navigation

StackViz

A visualization utility to help analyze the performance of DevStack setup and Tempest executions.

Installation

Installation of the frontend requires Node.js and Bower. On Ubuntu:

sudo apt-get install nodejs npm nodejs-legacy
sudo npm install -g bower

Then, install the Bower components by running, from the project directory:

bower install

Lastly, install the project. Pip is recommended, like so:

sudo pip install .

Usage - Server

First, install the necessary dependencies with Pip:

sudo pip install -r requirements.txt

The Django development server may then be used to view the interface. Run:

python manage.py runserver

You can then browse to the printed URL in your browser of choice.

Usage - Static Site

The server can be "snapshotted" and exported to a static HTML site using the installed stackviz-export utility. StackViz can then be viewed using any web browser with no requirement of any server-side processing.

To generate, run:

stackviz-export -r path/to/testrepository/ dest_dir

... where dest_dir is the path to a target directory where files should be written. When finished, the index.html file can be opened in a browser. Note that the above gathers test data from a testrepository directory, though direct subunit streams either from files or standard input are also supported. For more information, see stackviz-export --help.

Note that some browsers enforce content origin policies that may disallow XHRs when viewed directly from the local filesystem. To work around this, you can use something like the Python SimpleHTTPServer:

python -m SimpleHTTPServer

GZipped Data

As the log data can become quite large, exported files can be compressed with GZip to significantly reduce the size of the data files. To enable, run:

stackviz-export -r path/to/testrepository/ --gzip dest_dir

Data files will then be written in compressed form, and will be suffixed with *.json.gz. Note that web servers must be properly configured to serve pre-compressed files. Notably, Python's SimpleHTTPServer will not do this by default. However, Twisted can be used as a drop-in replacement as follows:

twistd -no web --path=.

Other web servers, such as Apache, should also serve these files correctly without any extra configuration.

(Specifically, the response must have headers Content-Type: application/json and Content-Encoding: gzip.)

DStat Data

StackViz will also show charts generated from DStat logs, if available. Note that console output from DStat is not sufficient - a CSV logfile must be used. Then, provide the logfile to stackviz-export:

stackviz-export -r testrepository/ --dstat path/to/dstat.csv dest_dir

Log Locations

Log locations are configured along with normal Django settings in stackviz/settings.py, or specified as command-line arguments to stackviz-export. Several different types of logs are rendered by StackViz are read by default from:

  • Tempest (testr repositories): ./test_data/
  • Dstat: ./dstat.log
  • DevStack: TODO

Testing

Server (Python)

Server-side tests can be run using Tox:

tox

A linter (flake8) will be run automatically and its output included in the report.

Client (JavaScript)

Client-side tests are run via Karma. To run, install the karma-cli and the npm dependencies:

npm install
sudo npm install --global karma-cli

Then, run Karma:

karma start --single-run

Tests will be executed using PhantomJS by default. Similarly, ESLint can be used to verify formatting:

eslint stackviz/static/

About

A performance visualization utility for DevStack and Tempest

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published