PerfKit Explorer Installation Guide
PerfKit is a service & web front end for composing queries and dashboards, and sharing the results. You can see a [live deployment and demo dashboard here] (https://perfkit-explorer.appspot.com/explore?dashboard=5714163003293696). You can click and interact with widgets (graphs), see the SQL, edit the json, and move things around on the demo dashboard. When you visit the site it will ask you permission to use your account. The full secuity model of the dashboard is active and permission is needed to verify your identity for permission checking. Everyone has read. Writing dashboards and execution of arbitrary queries is restricted.
Note the installation instructions below are based on using a Google Cloud Platform Compute instance, using the Debian Backports image. Instructions for platform installation may vary depending on your operating system and patch levels.
Set up your workstation
Install prerequisite packages:
sudo apt-get install python2.7 openjdk-7-jdk git nodejs nodejs-legacy npm
- Python 2.7 runtime
- Java 7 SDK
- NodeJS (nodejs-legacy provides the required /usr/bin/node symlink)
- Node Package Manager (NPM)
Install the Google Cloud SDK:
curl https://sdk.cloud.google.com | bash
- note: Restart your shell after installing gcloud to initialize the relevant paths.
Install the Google App Engine SDK for Python.
- note: You will need to add the App Engine SDK to your PATH so that you can find appcfg.py.
Create a root folder for your source code (i.e.:
~/projects) and navigate to it.
(Optional) If you are planning to participate in the open source project, create a GitHub account at http://www.github.com and make an editable copy of the PerfKitExplorer repository by clicking the "Fork" button at the top right. You can also set this up later, but it's easier to make the fork now before cloning it.
Clone the repository:
git clone https://github.com/GoogleCloudPlatform/PerfKitExplorer.git
or, if you created your own fork in the step above:
git clone https://github.com/MY_GITHUB_ID/PerfKitExplorer.git
If you have connected and authenticated correctly, the source code for PerfKit Explorer will download. Your projects folder will contain:
(projects) PerfKitExplorer bin client ... compile.sh app.yaml
Change to the PerfKitExplorer folder and download the Closure Tools, which are included as a submodule in the project:
git submodule update --init
Install the NPM packages for Gulp and dependencies, this will create a node_modules directory in the project.
Install Bower and the Bower packages for client-side dependencies, this will create a bower_components directory in the project.
npm install -g bower bower install
Create the App Engine project
Authorize your workstation to connect to Google Cloud resources:
gcloud auth login
Create a Google Cloud project at https://console.developers.google.com/project.
In "APIs & Auth > APIs", enable the BigQuery service.
- This is enabled by default for new projects.
Create the BigQuery repository
Create a Google Cloud project, or use the same one you used for the App Engine project.
Enable billing for your Cloud Project (available from links on the left-hand side) https://console.developers.google.com/project/MY_PROJECT_ID/settings
- MY_PROJECT_ID is the unique "Project ID", not the human-readable project name.
Create a dataset (ex: samples_mart):
bq mk --project=MY_PROJECT_ID samples_mart
Upload sample data to a new table in your dataset (ex: results):
pushd ~/projects/PerfKitExplorer/data/samples_mart bq load --project_id=MY_PROJECT_ID \ --source_format=NEWLINE_DELIMITED_JSON \ samples_mart.results \ ./sample_results.json \ ./results_table_schema.json popd
If the App Engine and BigQuery projects are separate, add the service account from your App Engine project as an authorized use of your BigQuery project.
Compile and Deploy PerfKit Explorer
Navigate to the PerfKitExplorer directory:
app.yamlfile so that the instance class is appropriate for your needs, and the application name matches the project id you created in the 'Create the App Engine project' step, and the version string is set appropriately. For example:
application: perfkit-explorer-demo version: beta instance_class: F2
config/data_source_config.jsonso that the production tags are appropriate for the repository you created in the previous step. For example:
project_id: perfkit-explorer-demo project_name: perfkit-samples samples-mart: perfkit-samples.samples_mart analytics-key: UA-12345
Compile the application.
You will now find a
Deploy PerfKit Explorer to App Engine.
appcfg.py --oauth2 update deploy
By default the application will be deployed to a build/version specific to your client. For example, with the following values:
application: MY_PROJECT_ID version: 15
will deploy to http://15-dot-MY_PROJECT_ID.appspot.com
Set up a PerfKit dashboard
Open the project URL http://15-dot-MY_PROJECT_ID.appspot.com in your browser.
Click "Edit Config" in the gear icon at the top right and set "default project" to the project id.
In Perfkit Dashboard Administration, click "Upload", and select the sample dashboard file: PerfKitExplorer/data/samples_mart/sample_dashboard.json