How to set multiple commands in one yaml file with Kubernetes?

In Kubernetes, if you need to execute multiple commands when a container in a Pod starts, there are typically two methods to achieve this:

Method 1: Using an Array Directly in YAML

In the Kubernetes YAML configuration file, you can directly specify a command array in the spec.containers.command field, where each element represents a command. However, note that typically each container can only start one main process, so you need to use a shell like sh or bash to execute multiple commands.

For example, the following is a Pod configuration example where the Pod first prints a message and then sleeps for 100 seconds:

yaml
apiVersion: v1
kind: Pod
metadata:
  name: multi-command-pod
spec:
  containers:
  - name: example-container
    image: ubuntu
    command: ["/bin/sh"]
    args: ["-c", "echo 'Hello, Kubernetes!'; sleep 100"]

In this example, the command specifies using /bin/sh to execute the command, and the -c element in the args array tells the shell to execute the command string that follows. The commands within the string are executed sequentially using semicolons.

Method 2: Using a Startup Script

Another approach is to write your commands into a script file and execute it when the container starts. First, create a script file containing all your commands, and add it to the Docker image during the build process.

bash
#!/bin/bash
# startup-script.sh
echo "Hello, Kubernetes!"
sleep 100

Then, add this script to the Dockerfile and set it as the ENTRYPOINT:

Dockerfile
FROM ubuntu
ADD startup-script.sh /usr/local/bin/startup-script.sh
RUN chmod +x /usr/local/bin/startup-script.sh
ENTRYPOINT ["/usr/local/bin/startup-script.sh"]

In this case, the Kubernetes YAML file does not need to specify the command or args fields:

yaml
apiVersion: v1
kind: Pod
metadata:
  name: multi-command-pod
spec:
  containers:
  - name: example-container
    image: custom-image-with-script  # Using an image containing the script

Both methods can execute multiple commands in a Kubernetes Pod. The choice depends on the specific scenario and personal preference. Using shell commands chained together provides a quick implementation, while using scripts makes command execution more modular and facilitates centralized management.