3

Follwing the documentation and provided example here: https://kubernetes.io/docs/concepts/workloads/controllers/jobs-run-to-completion/#running-an-example-job

I run kubectl apply -f job.yaml

kubectl apply -f job.yaml                                    
job.batch/pi created

Monitoring the job with get pods pi-fts6q 1/2 Running 0 52s

I always see 1/2 Running even after the job is complete and checking the logs shows it is completed.

How can I get the job to show a completed status? The job will stay in a running state showing no completions forever.

Parallelism:    1
Completions:    1
Start Time:     Thu, 06 Jun 2019 16:21:36 -0500
Pods Statuses:  1 Running / 0 Succeeded / 0 Failed

It seems the underlying pod that did the work completed, but the actual job-controller stays alive forever.

Cromat
  • 170
  • 1
  • 16
  • 2
    Found the answer, I'm running an istio proxy envoy: https://github.com/istio/istio/issues/6324 – Cromat Jun 06 '19 at 21:27

2 Answers2

1

Problem is in incomplete implementation of proxy-agent.

Simply add a '/quitquitquit' handler to proxy-agent.

Users can manually add a curl or http request to localhost to stop sidecar when the job is done. P0 is a workaround.

More information you can find here: istio-issue

Malgorzata
  • 6,409
  • 1
  • 10
  • 27
1

I just added these two curls at the end of the script of the job to kill istio and the pilot:

curl --request POST localhost:15000/quitquitquit || echo 'No istio it is then'
curl --request POST localhost:15020/quitquitquit || echo 'No istio at all, not even pilot'
leo
  • 145
  • 1
  • 3