Quisby
(Unoffical name) Quisby: An idler; one who does not or will not work. (noun)
Quisby is a tool to provide first view into the results from various benchmarks such as linpack, streams, fio etc. It doesn't aim to replace existing data viz tool but rather to provide a simplified view to the data with basic metric to understand the benchmark results from a higher level view. For detailed view, there are other tools such as pbench-dashboard, js-charts etc at hand.
Bechmarks currently supported:
| Benchmark | Source data |
|---|---|
| linpack | Benchmark result |
| streams | Summary result |
| uperf | Summary csv result |
| specjbb | Benchmark result |
| pig | Benchmark result |
| hammerDB | Benchmark result |
| fio | pbench result |
| autohpl | Summary result |
| aim | Benchmark result |
| etcd | pbench result |
| reboot | Benchmark result |
| speccpu | Benchmark result |
What it does
It extracts data from benchmark results file or summary results produced by wrapper benchmark programs and move that results to Google Sheet via sheets API V4.
Executable creation
pyinstaller --noconfirm --log-level=INFO -p quisby/ quisby.py
This command creates quisby executable. It binds all the dependencies into a file, making it easy to use.
Prerequisites to run quisby executable
- A google service account credentials(credentials.json) to hit various google APIs.
- Create a folder ~/.config/quisby/ if it doesn't exist.
- Copy credentials.json to the created folder.
- Copy example.ini file to ~/.config/quisby/config.ini.
test_name(Name of test)
test_path( Path to test results directory )
results_location( Path to results_location file )
system_name( Mention cloud/baremetal system )
spreadsheetId( Mention spreadsheet ID if exists, otherwise quisby creates a new one for you )
users( Mention the users you need to give access too. Example - abc@gmail.com,xyz@gmail.com )
How to run quisby
- Extract the tar to the executable
- Fill out the fields in config.ini file.
- Run ./quisby.app
Run quisby using pip
$ $ pip install quisby
test: results_linpack
</path/to/results>
...
test: pbench_fio
<http url with results>
...
Then you can run
$ quisby process --os-type <add-here> --os-release <add-here> --cloud-type <add-here> location_file`For more information on options, run:
$ quisby -hThat's it. It will return a google sheet. Visit the google sheet page and you will see a newly created spreadsheet with the data populated and graphed.
Comparison
If you want to compare two different OS release of similar system type then there are scripts that will help you to create a spreadsheet for the same.
and then run:
quisby compare --test-name <benchmark-name-(optional)> --spreadsheets <spreadsheet1,spreadsheet2>and it would return a newly created spreadsheet with the comparison data.
Development
#Clone the repo git clone git@github.com:sousinha1997/quisby.git
Installation
source ./install.sh (optional, for configuring aws and/or azure cli) source ./install -aws -azure
Contributing
Create issues and create a seperate feature branch to work on it. Push the changes to your clone repo and then create a pull request to the master branch of the origin repo.
