Docker Deployment VM s and Docker Executors
Docker executors run in a Docker deployment VM. Like a VM build executor, a Docker deployment VM is an OCI VM Compute instance dedicated to run builds.
To run builds in Docker executors, you add Docker deployment VMs. A Docker executor can run only one build at a time, but a Docker deployment VM can run multiple builds simultaneously, depending on how many Docker executors are configured for the deployment VM. The maximum number of Docker executors that can be run by a given Docker deployment VM is configurable between 1 and 10.
When a job’s build is triggered, VB Studio starts a Docker deployment VM if it is in a stopped state, creates a Docker executor in it, and builds the job using the executor.
Remember, a Docker deployment VM costs you only when it’s in a running state and actively building jobs. When the Docker deployment VM is idle and not building jobs, it’ll be put to sleep. You can configure the number of minutes the deployment VM remains idle before it is put to sleep.
This table describes the Docker deployment VM states:
| State | What does it mean? | Does it cost? |
|---|---|---|
| Pending | After you add a Docker deployment VM, it is in this state until the management VM finishes building, and then it will start automatically. | No |
| Starting | The Docker deployment VM is starting. If the Docker deployment VM starts from the Pending state, VB Studio installs the operating system and software packages on the Docker deployment VM’s assigned boot volume. This takes time. If the Docker deployment VM starts from the Stopped state, VB Studio uses software packages and the operating system from the previous run’s saved boot volume. |
Yes |
| Ready | The operating system and Docker agent are installed and started. The VM is ready to deploy Docker executors. | Yes |
| Stopping | The Docker deployment VM is shutting down. Before shutting the Docker deployment VM down, VB Studio saves the operating system and software packages to the Docker deployment VM’s assigned boot volume. |
Yes |
| Stopped | The Docker deployment VM has shut down. | No |
| Error | There’s a hardware or a software issue on the Docker deployment VM. Check the Docker deployment VM’s log to find more about the cause. | No |
Some key points to remember about Docker deployment VMs and Docker executors:
- Once created, the Docker Management VM is always running and costs you 24-7. To keep your costs lower, if you are planning to run fewer builds with fewer images, select a smaller shape with fewer OCPUs.
- To run builds in Docker executors, you add Docker deployment VMs and specify the number of Docker executors it can run, select the OCI region from the connected OCI account’s subscribed regions, the OCI Compute VM’s shape, and optionally the VCN.
- Your OCI account may have some Compute instance limits set. When you add a Docker deployment VM, VB Studio looks into the specified OCI region’s Availability Domains, finds available OCPUs with the specified shape, calculates the number of Compute instances, and displays the number of Compute VM instances you can add from your OCI account’s set limit.
- When you add your first Docker deployment VM, VB Studio adds one more VM, called the Docker Management VM, to manage Docker images.
- The Docker Management VM is responsible for creating Docker images and Docker executors and for deploying images to Docker executors.
- When you create your first Docker deployment VM, the Docker Management VM is also created with the same shape.
- VB Studio provides a default shape selection for the management VM. This default is only a recommendation. Some things to consider:
- You could still choose a lower-end configuration for your specific purposes, but the default is what VB Studio determines to be the optimal powered VM for the management VM. Although a low-end VM (such as Standard 1.1 with 1 CPU and 8 GB of memory) could be used for Docker Image management in scenarios that have few Executor templates and create just a few Docker images, it is better to use the default recommendation. If you try to select a lower-powered VM than what is recommended, you’ll see a warning.
- For scenarios that have several Executor templates and will create several Docker images, using a high-end VM (such as VM.Standard.E3.Flex with 4 OCPU and 32 GB RAM) instead is recommended. Using a low-end VM in demanding scenarios will severely hamper overall performance.
- If a Docker deployment VM is not in use, it will eventually time out and enter the Stopped state, which has no associated cost to you.
- The higher the number of Docker deployment VMs running Docker executors at any specific time, the higher the associated cost. To minimize the higher cost, you should configure the sleep timeout to stop inactive Docker deployment VMs after some period of time. The default timeout is 300 minutes. The sleep timeout setting will be applied to all your organization’s Docker deployment VMs.