Get Kubernetes Job Logs

When using Kubernetes, often you need to read logs from the running pods.

On a bigger cluster you may have many pods that are running and many that have either completed or errored out.

When deploying Kubernetes Jobs, often you want to read the logs of the job, whether it completed or has any errors.

Finding the correct pod amongst many pods may be a bit of a challenge.

In this post you’ll learn how to list the correct job pods - all of them, and how to retrieve the job logs - from the current

completed pod and from a pods that exited with error.

Given you already have a Kubernetes cluster, and the job was deployed in a namespace, this is what you want to do.

List the job pods

Let’s say we have a job like this:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
apiVersion: batch/v1

kind: Job

metadata:

  name: calculate-fibonacci

  namespace: example

spec:

  template:

    spec:

      containers:

      - name: fibonacci

        image: node:16-alpine

        command: ['node',  '-e', 'function f(x){ if(x === 1 || x === 2) return 1; return f(x-1)+f(x-2);} console.log("The 10th fibonacci number is: ", f(10));']

      restartPolicy: Never

  backoffLimit: 4

The Job definition:

• The name of the job is calculate-fibonacci

• The job is in namespace called example

To list the pods associated with this job, do:

1
2
3
kubectl -n example get pods -l job-name=calculate-fibonacci

• -n example - this tells kubectl to work in the example namespace

• get pods will give us the list of pods in the namespace

• -l job-name=calculate-fibonacci this refines the list of pods that have a label job-name with the name of our job: calculate-fibonacci

This will produce the following output:

1
2
3
4
5
NAME                           READY   STATUS      RESTARTS   AGE

calculate-fibonacci--1-pdzg4   0/1     Completed   0          3s

Get the job logs

Once we know the pods associated with a job, we can view the logs the same way we view the logs for any pod: kubectl logs <the id of the pod>:

1
2
3
kubectl -n example logs calculate-fibonacci--1-pdzg4

It will give us the logs for the job:

1
2
3
The 10th fibonacci number is:  55

If the job is still running and we want to follow the logs as they are printed out, just add -f (follow) to the command:

1
2
3
kubectl -n example logs -f calculate-fibonacci--1-pdzg4

One-liner for viewing the logs of a completed job

Now we can combine the both commands into a single one, to view the logs for a completed job:

1
2
3
kubectl -n example logs $(kubectl -n example get pods -l job-name=calculate-fibonacci --field-selector='status.phase=Succeeded' --no-headers -o custom-columns=':metadata.name')

It will give us the logs of the completed pod directly:

1
2
3
The 10th fibonacci number is:  55

Notice that we added a couple of things in the command that selects the pod:

• --field-selector='status.phase=Succeeded' - we want to find the pod that has Succeeded status (everything went OK)

• --no-headers - we don’t want any headers in the output, as we only want the pod id

• custom-columns=':metadata.name' - show just the metadata.name of the pod (the id)