Helm Job timeout

General Discussions
1

Hi,

I am investigating a method of running database migrations (Ruby on Rails, Django, etc) in Helm in a Job with “helm.sh/hook”: pre-install,pre-upgrade . By all accounts, it should work fine. This seems like the best way to run database migrations via a helm deployment.

However, in a blog post someone mentioned that if the Job takes too long it might time out.

Would Helm’s global 5 minute timeout apply to pre-install jobs? That is, if a Job takes more than 5 minutes, for whatever reason, any reason, even if it’s constantly returning data to stdout/stderr, then Helm will consider that the Job has “timed out”? What are the criteria for a timeout to happen on a pre-install Job in Helm?

Thanks.

2

Hello @sdarwin did you find a solution, I been trying to run Django helm chart with pre-install,pre-upgrade jobs but keep getting error

I will appreciate if you or any member could share your solution

3

Hi Stephen,

You didn’t mention the errors you were getting. This thread was a general question about timeouts. Otherwise migrations were working. Here’s an example, it would have to be modified for your environment.

apiVersion: batch/v1
kind: Job
metadata:
  name: database-migrations-job
  labels:
    env: {{ .Values.deploymentEnvironment }}
    app: boost
  annotations:
    "helm.sh/hook": pre-install,pre-upgrade
    "helm.sh/hook-weight": "-1"
spec:
  ttlSecondsAfterFinished: 3600
  backoffLimit: 4
  template:
    spec:
      restartPolicy: OnFailure
      volumes:
{{ toYaml .Values.Volumes | indent 6 }}
      containers:
        - name: database-migrations
          image: {{ .Values.Image }}:{{ .Values.ImageTag }}
          command: ["/bin/sh", "-c"]
          args:
            - "./manage.py migrate --noinput"
          env:
{{ toYaml .Values.Env | indent 10 }}
          volumeMounts:
{{ toYaml .Values.VolumeMounts | indent 10 }}