Skip to content

Commit

Permalink
Experimental: Selenium Grid scaler in K8s implementation preview (#2400)
Browse files Browse the repository at this point in the history
  • Loading branch information
VietND96 committed Sep 19, 2024
1 parent e27f6ed commit 2af6166
Show file tree
Hide file tree
Showing 15 changed files with 2,555 additions and 21 deletions.
4 changes: 3 additions & 1 deletion .github/workflows/deploy.yml
Original file line number Diff line number Diff line change
Expand Up @@ -152,7 +152,9 @@ jobs:
continue_on_error: true
command: VERSION="${GRID_VERSION}" BUILD_DATE=${BUILD_DATE} make release_latest
- name: Update package versions
run: make generate_latest_sbom
run: |
make generate_latest_sbom
make fetch_grid_scaler_resources
- name: Tag browser images
if: github.event.inputs.skip-build-push-image != 'true'
uses: nick-invision/retry@master
Expand Down
62 changes: 62 additions & 0 deletions .keda/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,62 @@
# Introduction

Selenium Grid Scaler is a built-in scaler is maintained in upstream KEDA [repository](https://github.com/kedacore/keda). The scaler implementation could be found [here](https://github.com/kedacore/keda/blob/main/pkg/scalers/selenium_grid_scaler.go). The official docs of the scaler could be seen [here](https://keda.sh/docs/latest/scalers/selenium-grid-scaler/).

Now, [SeleniumHQ/docker-selenium](https://github.com/SeleniumHQ/docker-selenium) involves as the maintainer for the scaler.

In order to deliver and get feedback continuously on any new bug fixes, improvement, or features for the Selenium Grid scaler. We select the latest stable version of KEDA core, patch the scaler implementation then build and deploy KEDA container images following our image tag convention.

The stable implementation will be merged to the upstream KEDA repository frequently and will be available in the next KEDA core release.

# How to use the patched scaler

Replace the image registry and tag of these KEDA components with the patched image tag:

```bash
docker pull selenium/keda:2.15.1-selenium-grid-20240907
docker pull selenium/keda-metrics-apiserver:2.15.1-selenium-grid-20240907
docker pull selenium/keda-admission-webhooks:2.15.1-selenium-grid-20240907
```

Besides that, you also can use image tag `latest` or `nightly`.

If you are deploying KEDA core using their official Helm chart, you can overwrite the image registry and tag by providing the following values in the `values.yaml` file. For example:

```yaml
image:
keda:
registry: selenium
repository: keda
tag: "2.15.1-selenium-grid-20240907"
metricsApiServer:
registry: selenium
repository: keda-metrics-apiserver
tag: "2.15.1-selenium-grid-20240907"
webhooks:
registry: selenium
repository: keda-admission-webhooks
tag: "2.15.1-selenium-grid-20240907"
```
If you are deployment Selenium Grid chart with `autoscaling.enabled` is `true` (implies installing KEDA sub-chart), KEDA images registry and tag already set in the `values.yaml`. Refer to list [configuration](../charts/selenium-grid/CONFIGURATION.md).

# Pull requests under testing

Here is list of pull requests that are under testing and will be merged to the upstream KEDA repository.
You can involve to review and discuss the pull requests to help us early detect and fix any issues.

[kedacore/keda](https://github.com/kedacore/keda)

- https://github.com/kedacore/keda/pull/6169

[kedacore/keda-docs](https://github.com/kedacore/keda-docs)

- https://github.com/kedacore/keda-docs/pull/1468

# Resources

You can inspect the implementation of current Selenium Grid scaler:

- [selenium_grid_scaler.go](./scalers/selenium_grid_scaler.go)
- [selenium_grid_scaler_test.go](./scalers/selenium_grid_scaler_test.go)
- [selenium-grid-scaler.md](./scalers/selenium-grid-scaler.md)
207 changes: 207 additions & 0 deletions .keda/scalers/selenium-grid-scaler.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,207 @@
+++
title = "Selenium Grid Scaler"
availability = "v2.4+"
maintainer = "Volvo Cars, SeleniumHQ"
category = "Testing"
description = "Scales Selenium browser nodes based on number of requests waiting in session queue"
go_file = "selenium_grid_scaler"
+++

### Trigger Specification

This specification describes the `selenium-grid` trigger that scales browser nodes based on number of requests in session queue and the max sessions per grid.

The scaler creates one browser node per pending request in session queue, divided by the max amount of sessions that can run in parallel. You will have to create one trigger per browser capability that you would like to support in your Selenium Grid.

The below is an example trigger configuration for chrome node.

```yaml
triggers:
- type: selenium-grid
metadata:
url: 'http://selenium-hub:4444/graphql' # Required. Can be ommitted if specified via TriggerAuthentication/ClusterTriggerAuthentication.
browserName: 'chrome' # Required
browserVersion: '91.0' # Optional. Only required when supporting multiple versions of browser in your Selenium Grid.
unsafeSsl : 'true' # Optional
activationThreshold: 5 # Optional
platformName: 'Linux' # Optional
```
**Parameter list:**
- `url` - Graphql url of your Selenium Grid. Refer to the Selenium Grid's documentation [here](https://www.selenium.dev/documentation/en/grid/grid_4/graphql_support/) to for more info.
- `username` - Username for basic authentication in GraphQL endpoint instead of embedding in the URL. (Optional)
- `password` - Password for basic authentication in GraphQL endpoint instead of embedding in the URL. (Optional)
- `browserName` - Name of browser that usually gets passed in the browser capability. Refer to the [Selenium Grid's](https://www.selenium.dev/documentation/en/getting_started_with_webdriver/browsers/) and [WebdriverIO's](https://webdriver.io/docs/options/#capabilities) documentation for more info.
- `sessionBrowserName` - Name of the browser when it is an active session, only set if `BrowserName` changes between the queue and the active session. See the Edge example below for further detail. (Optional)
- `browserVersion` - Version of browser that usually gets passed in the browser capability. Refer to the [Selenium Grid's](https://www.selenium.dev/documentation/en/getting_started_with_webdriver/browsers/) and [WebdriverIO's](https://webdriver.io/docs/options/#capabilities) documentation for more info. (Optional)
- `unsafeSsl` - Skip certificate validation when connecting over HTTPS. (Values: `true`, `false`, Default: `false`, Optional)
- `activationThreshold` - Target value for activating the scaler. Learn more about activation [here](./../concepts/scaling-deployments.md#activating-and-scaling-thresholds). (Default: `0`, Optional)
- `platformName` - Name of the browser platform. Refer to the [Selenium Grid's](https://www.selenium.dev/documentation/en/getting_started_with_webdriver/browsers/) and [WebdriverIO's](https://webdriver.io/docs/options/#capabilities) documentation for more info. (Default: `Linux`, Optional)
- `nodeMaxSessions` - Number of maximum sessions that can run in parallel on a Node. (Default: `1`, Optional). Update this parameter align with node config `--max-sessions` (`SE_NODE_MAX_SESSIONS`) to have the correct scaling behavior.

### Example

Here is a full example of scaled object definition using Selenium Grid trigger:

```yaml
apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
name: selenium-grid-chrome-scaledobject
namespace: keda
labels:
deploymentName: selenium-chrome-node
spec:
maxReplicaCount: 8
scaleTargetRef:
name: selenium-chrome-node
triggers:
- type: selenium-grid
metadata:
url: 'http://selenium-hub:4444/graphql'
browserName: 'chrome'
```

The above example will create Chrome browser nodes equal to the requests pending in session queue for Chrome browser.

Similarly for Firefox

```yaml
apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
name: selenium-grid-firefox-scaledobject
namespace: keda
labels:
deploymentName: selenium-firefox-node
spec:
maxReplicaCount: 8
scaleTargetRef:
name: selenium-firefox-node
triggers:
- type: selenium-grid
metadata:
url: 'http://selenium-hub:4444/graphql'
browserName: 'firefox'
```

Similarly for Edge. Note that for Edge you must set the `sessionBrowserName` to `msedge` inorder for scaling to work properly.

```yaml
apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
name: selenium-grid-edge-scaledobject
namespace: keda
labels:
deploymentName: selenium-edge-node
spec:
maxReplicaCount: 8
scaleTargetRef:
name: selenium-edge-node
triggers:
- type: selenium-grid
metadata:
url: 'http://selenium-hub:4444/graphql'
browserName: 'MicrosoftEdge'
sessionBrowserName: 'msedge'
```

If you are supporting multiple versions of browser capability in your Selenium Grid, You should create one scaler for every browser version and pass the `browserVersion` in the metadata.

```yaml
apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
name: selenium-grid-chrome-91-scaledobject
namespace: keda
labels:
deploymentName: selenium-chrome-node-91
spec:
maxReplicaCount: 8
scaleTargetRef:
name: selenium-chrome-node-91
triggers:
- type: selenium-grid
metadata:
url: 'http://selenium-hub:4444/graphql'
browserName: 'chrome'
browserVersion: '91.0'
```

```yaml
apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
name: selenium-grid-chrome-90-scaledobject
namespace: keda
labels:
deploymentName: selenium-chrome-node-90
spec:
maxReplicaCount: 8
scaleTargetRef:
name: selenium-chrome-node-90
triggers:
- type: selenium-grid
metadata:
url: 'http://selenium-hub:4444/graphql'
browserName: 'chrome'
browserVersion: '90.0'
```

### Authentication Parameters

It is possible to specify the Graphql url of your Selenium Grid using authentication parameters. This useful if you have enabled Selenium Grid's Basic HTTP Authentication and would like to keep your credentials secure.

- `url` - Graphql url of your Selenium Grid. Refer to the Selenium Grid's documentation [here](https://www.selenium.dev/documentation/en/grid/grid_4/graphql_support/) for more info.
- `username` - Username for basic authentication in GraphQL endpoint instead of embedding in the URL. (Optional)
- `password` - Password for basic authentication in GraphQL endpoint instead of embedding in the URL. (Optional)

```yaml
apiVersion: v1
kind: Secret
metadata:
name: selenium-grid-secret
namespace: keda
type: Opaque
data:
graphql-url: base64 encoded value of GraphQL URL
graphql-username: base64 encoded value of GraphQL Username
graphql-password: base64 encoded value of GraphQL Password
---
apiVersion: keda.sh/v1alpha1
kind: TriggerAuthentication
metadata:
name: keda-trigger-auth-selenium-grid-secret
namespace: keda
spec:
secretTargetRef:
- parameter: url
name: selenium-grid-secret
key: graphql-url
- parameter: username
name: selenium-grid-secret
key: graphql-username
- parameter: password
name: selenium-grid-secret
key: graphql-password
---
apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
name: selenium-grid-chrome-scaledobject
namespace: keda
labels:
deploymentName: selenium-chrome-node
spec:
maxReplicaCount: 8
scaleTargetRef:
name: selenium-chrome-node
triggers:
- type: selenium-grid
metadata:
browserName: 'chrome'
authenticationRef:
name: keda-trigger-auth-selenium-grid-secret
```
Loading

0 comments on commit 2af6166

Please sign in to comment.