Dynamic Worker pools provide a quick and easy way to use workers for your deployments. They are a special type of worker pool available on Octopus Cloud.
Dynamic workers are isolated virtual machines, created on-demand to run your deployments and runbook steps. They run on Azure and are created and managed by Octopus Cloud, which means you don’t need to configure or maintain additional infrastructure.
On demand
A dynamic worker is created on demand and leased to an Octopus Cloud instance for a limited time before being destroyed.
Octopus Cloud will automatically destroy dynamic workers as soon as one of these conditions is met:
- The worker has been idle for 60 minutes.
- The worker has existed for 72 hours (3 days).
Please reach out to our support team if you need these values to be adjusted for your instance.
Worker VMs are provisioned with at least 20GB of available disk space, which is persistent until the worker is destroyed.
Isolated
Each worker VM is provisioned exclusively to a specific customer, and is completely isolated from other customers.
Dynamic Worker Images
Each dynamic worker pool uses a specific worker image. This is a VM image which determines the operating system (and OS version) running on the worker. You can edit a dynamic worker pool to change which image is used.
When you sign up to Octopus Cloud (or create a new space), you automatically receive a worker pool for the Ubuntu (default) image, and a worker pool for the Windows (default) image.
The full list of available worker images includes both specific operating system versions (e.g., Ubuntu Linux 22.04), and also generic “default” options such as Ubuntu (default). Choosing the default option means that your worker will use the latest stable worker image when it is released.
The current default images are:
Ubuntu (default)➜Ubuntu Linux 22.04Windows (default)➜Windows Server Core 2022
Choosing an Image
The default image is a good option to choose if you are:
- Running a simple script that doesn’t require specific tools or operating system versions
- Running a step inside a container
If you’re writing a script that relies on a specific version of tooling (e.g., Helm), then we recommend using execution containers for workers to run the script in a Docker container with the tool versions you need.
Alternatively, you can choose a specific worker image, instead of the “default” options, to prevent worker image upgrades from impacting your deployments.
| Type | Pros | Cons |
|---|---|---|
Default (eg Ubuntu (default)) | Automatically uses the latest image. Deployments will continue to work even when a worker image is marked as deprecated or decommissioned. | The versions of dependencies (e.g. Helm) are not fixed. Deployments that rely on specific versions of dependencies or operating system specific features may break during upgrades. |
Specific (e.g., Ubuntu Linux 22.04) | The version of the operating system and dependencies are fixed and can be relied upon. | When a worker image is marked as deprecated, warnings will start to appear in your deployment logs. When a worker image is decommissioned, you will need to take action to update your worker pool or deployments will fail. |
Deprecation
When an image is marked as deprecated, you will see warnings in the Octopus UI, and in the deployment log. After a suitable deprecation period, deployments will start to fail if they target an image that has hit end-of-life.
When you start getting warnings in your deployments and/or see deprecation warnings in the Octopus portal, please plan to modify your worker pool to use a different image and test your scripts on the new image.
If your Worker Pool is set to use the Operating System default, for example, Ubuntu (default), the default will be swapped over to a new Operating System version by Octopus Deploy. Your deployments and runbooks will automatically use the new version.
You should validate that your deployments and runbooks work with the new version prior to the cutover date. The new image will be made available prior to the cutover date and we will notify you of the cutover date to give you time to undertaking any required testing.
Modifying the worker pool
If the worker pool has been configured to specifically use a deprecated worker type, you will need to update the worker image on the worker pool.
Navigate to your space, then go to Infrastructure ➜ Worker Pools. Worker pools with a deprecated worker image will show a Deprecated label next to the worker pool.
Click the overflow menu (...) and click Edit. You can then select a new worker image from the dropdown list, such as Ubuntu (default) or a specific operating system version.
Available Dynamic Worker Images
Worker images are rebuilt on a regular basis, so that the operating system is up to date with the latest security patches.
Ubuntu 22.04
This is the default for the Ubuntu operating system, referenced as Ubuntu (default).
Each Ubuntu Server 22.04 worker is provisioned with a baseline of tools including (but not limited to):
- .NET 6
- Docker (latest)
- Powershell Core (latest)
- Python 3 (latest)
- GCloud CLI (367.0.0)
Ubuntu workers are designed to use execution worker containers for tooling such as kubectl and helm. This makes it much easier to choose the appropriate runtime environment with the tools you need for your use case.
Windows Server Core 2022
This is the default for the Windows operating system, referenced as Windows (default).
Each Windows Server Core 2022 worker is provisioned with a baseline of tools including (but not limited to):
- .NET Core (2.1, 3.1)
- .NET Framework 3.5
- .NET Framework 4.8
- AWS IAM Authenticator (0.5.3)
- Chocolatey (latest)
- Docker (latest)
- Helm (2.9.1)
- Kubectl (multiple versions)
- Microsoft Service Fabric (6.1.480.9494)
- Microsoft Service Fabric SDK (3.0.480)
- Nuget CLI (latest)
- Octopus Client (latest)
- Pip (latest)
- Powershell Core (latest)
- Python (3.7.4)
- GCloud CLI (339.0.0)
Windows 2022 workers are capable of running execution worker containers.
We recommend execution containers as the preferred option for steps requiring external tools. This allows you to control which version of the tools will be used as your scripts will rely on a specific version that they are compatible with to function correctly.
kubectl on Windows Images
Windows dynamic worker images come with many versions of kubectl available.
A specific version can be used by specifying a custom kubectl location of c:\tools\kubectl\{{version}}\kubectl.exe, where {{version}} is one of the following:
1.11.11.11.31.12.11.13.121.14.91.15.61.16.101.17.51.18.01.19.91.20.51.21.91.22.6
Installing Software On Dynamic Workers
Octopus does not recommend installing additional software on Dynamic Workers.
By default, every dynamic worker is destroyed after it has been idle for 60 minutes or allocated for over 72 hours. In addition Octopus cannot guarantee that the dynamic worker leased to run one step will be the same worker leased to other executing steps in a deployment or runbook run.
For deployments and runbook runs that require additional software dependencies on a dynamic worker, our recommendation is to leverage execution containers for workers. Octopus provides execution containers with a baseline of tools (octopusdeploy/worker-tools) pre-installed. These tools won’t include every possible software combination you might need. If you require a specific set of software and tooling we recommend building your own custom docker images for use with execution containers.
Octopus worker-tools are cached on Dynamic Workers
The octopusdeploy/worker-tools images provided for the execution containers feature cache the five latest Ubuntu and two latest Windows Worker Tools images on a dynamic worker when it’s created. This makes them an excellent choice over installing additional software on a dynamic worker.
If you choose to install additional software on a dynamic worker, you are responsible for:
- Ensuring that software is installed at the start of each deployment or runbook run.
- Writing the necessary scripts to download and install that software.
- Verifying the latest version of the software works with the latest security patches of the host OS.
- Handling any issues that arise if a different dynamic worker is leased to different steps in your deployment or runbook run.
Learn more
Help us continuously improve
Please let us know if you have any feedback about this page.
Page updated on Tuesday, September 24, 2024