Jobs

Jobs

Schedule pods as Jobs rather than with a ReplicationController / ReplicaSet

With a ReplicatiionController a pod will b indefinitely restart if the pod stops

With a job resource, pods are expected to run a specific task and then exit.

3 main types of job:

1. Non Parell Jobs
2. Parallel jobs with a fixed completion count
3. Parallel Jobs with a work queue.

-------------------------------------

1. Non parallel jobs:
   The resource will monitor the job, and restart the pod if it fails or
   gets deleted
    - you can influence this with a restartPolicy attribute
   When the pod successfully completed, the job itself will be completed.

e.g.
apiVersion: batch/v1
kind: Job
metadata:
  name: pi
spec:
  template:
    metadata:
      name: pi
    spec:
      containers:
      - name: pi
        image: perl
        command: ["perl",  "-Mbignum=bpi", "-wle", "print bpi(2000)"]
      restartPolicy: Never

2. Parallel jobs with a fixed completion count:
  - In this case a Job can run multiple pods in parallel
  - You specify a fixed completion count
  - The job is complete when completion count == successful exited pods
  - To use: add "completions: " to the specification of your job.

3. Parallel jobs with a work queue:
   - In this case the pods should coordinate with themselves
     (or an external service) to determine what each should work on.
   - When any pod terminates with success, no more pods are created and the
     job is considered completed.
      This is because the pods should know themselves when all the work is done
       (because they communicate with each other) at that point when a pod terminates with success,
        then all pods should terminate because the job is completed.

helpful commands

...