Running a fleet

Follow these steps to run a fleet with the CLI or with the console. Note that fleets run as soon as they are created, so that running a fleet and creating a fleet are the same operation.

Before you begin

To successfully run a fleet, make sure you have followed the steps in Preparing to run your fleet. You also have the option to use Code Engine to build the image you reference when you create your fleet.

Running a fleet with the CLI

Use the following example command to run a fleet. This command allows additional options. For a complete list of all available command options, see the CLI docs.

ibmcloud code-engine fleet run --name my-fleet --image icr.io/codeengine/hello --tasks 1

--name: The name of the fleet. This value must be unique across all fleets within the same project. If no fleet name is specified, a random one is generated.
--image: The container image to use for the fleet.
--tasks: The number of tasks to run.

Example output.

Preparing your tasks: Please wait…
Launching fleet `my-fleet`…
Current fleet status `Launching`…
OK

Running a fleet with the console

Follow these steps to run a fleet in the Code Engine console.

Open the Code Engine console.
Click Start Creating.
Select a project from the list of available projects. You can also create a new one. You must have a selected project to run a fleet.
Select the option to create a fleet.
Specify a name for the fleet. Make sure the name is unique across all fleets within the project.
Specify a container image for your fleet, for example, icr.io/codeengine/helloworld. If you have your own source code that you want to turn into a container image, see Planning your build. For more information about the code that is used for this example, see helloworld.
Follow the prompts to configure your fleet.
In the Tasks section, configure the task specification method for the fleet. For more information, see Task specification.
In the Resources and scaling section, configure your instance resources and specify how your instances scale up or down.
In the Network placement section, review the subnets that your worker nodes deploy on.
In the Environmental variables and Volume mounts sections, add optional key-value pairs, configmaps, or additional files that can be used by your running code.
Click create.

Task specification

Fleets always have at least one task, but can have a much larger number. You can specify the number of tasks to complete, the number of tasks and instances to run at a time, and the order in which tasks are completed. When you run your fleet, worker nodes automatically scale up based on the number of tasks to be completed.

You can specify tasks in the following ways.

Number of tasks: To run a specific number of tasks, you can specify any positive integer. In the CLI, use the --tasks option. In the UI, select Number of tasks and enter the number you want to run.
Task index: To specify a range of tasks, you can specify a task index that includes a comma separated list of ranges and positive integers, such as 2-5,7-8,10. In the CLI, use the --task-indexes option. In the UI, select Task indexes and enter the range.
Task file: To have different tasks run with different commands or arguments, you can create a task specification file that overrides the image definition. This file should be formatted in JSON. In the CLI, specify the file with the OPTION option. In the UI, select Tasks from file and specify the file name. See an example of lines that can be added to a task specification file below.

Example lines in a task specification file.

{ "cmds": ["my", "multipart", "command"], "args": ["arg1", "arg2"]}
{ "cmds": ["other", "cmd"], "args": ["arg"]}
{ "args": ["argA", "argB", "argC"]}
{ "cmds": ["just", "another", "command"]}