Rancher has added a new feature in 1.4 for webhooks, with an initial driver to handle scaling. A key concept for implementing webhooks is that of a ‘Receiver’, which lets you register a webhook and provides a URL used to trigger an action inside of Rancher.
We have implemented webhooks with our new microservice, called webhook-service. I will explain the feature using our current driver, scaleService. The scaleService driver allows users to create a receiver hook for scaling a service up or down. A classic use case for this is integrating with a monitoring system to watch load balancer traffic, and calls this receiver hook when response times or requests per second reach a certain threshold. I will go through the steps of creating the receiver hooks in Rancher.
Adding a Receiver Hook
Navigate to the API -> Webhooks in the UI. This is where all the receiver hooks you create for the selected environment will be listed, under Receiver Hooks
Click on Add Receiver. You will see fields to be entered:
On this page, fields to be entered are as follows:
Name: Every receiver hook should be given a unique name so it can be easily identified
Kind: The Kind dropdown gives a list of all drivers available in webhook-service. Select a driver from this list (only Scale a Service is available for v1.4 – more drivers based on community input and requests to come!)
Action: This field lets you specify the action specific to the driver. For the scaleService driver, the only two actions are up (to increase the number of containers) and down (to decrease the number of containers).
The fields after this are specific to the scaleService driver:
Target Service: Select the service to be scaled from this dropdown (below)
By: This field asks the user to enter the amount by which the selected service should be scaled
Minimum Scale and Maximum Scale: The minimum and maximum number of containers your service can be allowed to have. Scaling the service beyond these values using webhooks won’t be allowed.
Click on Create once all fields are entered:
The created receiver hook will now be listed on the webhooks page. The Trigger URL is used to actually execute scaling up and down. Next to this field, there’s also options for deleting or cloning a receiver hook:
Using the Receiver Hook
The Trigger URL will execute the action for the driver. To use the trigger URL, you will need to do a POST to the specific URL. There is no authentication or body needed to POST to the URL. This URL consists of an API endpoint and a token. The token contains all the information for the created receiver hook, such as selected driver, service, action (up/down), and amount that the service should be scaled by. The API endpoint uses this token to make Cattle API calls and complete the action specified in the created receiver hook.
Using webhook-service for autoscaling
The scaleService driver from webhook-service can be used to autoscale a service based on container metrics. To do so, the webhook service must be used with an external monitoring service. Once you create a service, add a webhook for it, using the above steps. Then copy the Trigger URL and to use in the monitoring service: send a POST request to this trigger url when a particular metric, such as response time, crosses a specified threshold. Monitoring services such as Prometheus have a field called webhook_configs to specify such URLs that need a POST request sent to them. Once you have this configured, scaling the service will be handled by webhook-service.
Upcoming Drivers and Contributions
We plan on adding more drivers to webhook-service. For future releases, we are working on adding drivers for scaling up/down hosts and redeploying services based on DockerHub webhooks. We would also love to get contributions for new drivers as well. You can refer to these steps for adding new drivers and submit pull requests to our webhook-service repository.
Learn more about using webhooks for scaling – and more tips and tricks for running containers – in our upcoming online meetup, Tuesday, February 28. Register online here.
Join us for free online training courses, hosted monthly by a Rancher technical expert. We provide a great hands-on overview for new users setting up a Rancher deployment, and answer any and all questions you have about Rancher and how to integrate it into your DevOps processes!