K
Q

equivalent to `kubectl delete pods --all -n namespace` in helm

9/12/2019

I need to simply recreate all the pods on certain releases, say 10 in number, on a production cluster. Is there a helm command to loop over release names and pods under each release should be deleted and then recreated. any hints?

I am familiar with

for nmspc in $namespace_list; do kubectl delete pods -n $nmspc; done

However, I am trying to see if there is an equivalent helm command to do the same, such that once pods are deleted under the releases, they are spun up again.

for rlsz in $release_list; do helm delete pods -n $rlsz; done

Does something like this exist?

-- AhmFM
kubectl
kubernetes
kubernetes-helm

Similar Questions

How to run a Spark Standalone master on Kubernetes that will use the Kubernetes Cluser Manager to start workers
Creating an openshift cluster aws failure
How to tell helm to not create/change resource if it already exists?
OpenShift Web UI on upstream k8s

2 Answers

9/12/2019

You can try

helm upgrade --recreate-pods <release_name> path/to/chart

If your charts are organized in a list of directories, and directory name corresponds to release name (recommended approach), you can use

for d in */ ; do
    helm upgrade --recreate-pods "${d%?} "${d}"
done

where ${d%?} removes trailing slash from directory name.

Note, that by definition, recreate-pods will cause downtime.

If you simply want to restart the pods, and they are all deployed via Helm, you can simply do

kubectl delete pods -n my-namespace --all

Tiller will re-create them, as if they crashed.

-- Alex Pakka
Source: StackOverflow
9/13/2019

@Alex Pakka suggested you right approach with helm upgrade --recreate-pods <release_name> path/to/chart, but yeah sometimes it depends on chart.

Just for example I took nginx-ingress-1.15.0.tgz from stable/nginx-ingress

helm fetch stable/nginx-ingress

and installed in standard way

helm install --name nginx-ingress --namespace kube-system nginx-ingress-1.15.0.tgz

After deploying I was able successfully recreate pods with the below command:

helm upgrade --recreate-pods nginx-ingress nginx-ingress-1.15.0.tgz --username "xxxxx" --password "xxxxx"

---
nginx-ingress-controller-cd996946d-95ndx                       1/1     Running   0          47s
nginx-ingress-default-backend-77c7c664bb-2nsdp                 1/1     Running   0          46s
---
nginx-ingress-controller-cd996946d-95ndx                       1/1     Terminating         0          93s
nginx-ingress-controller-cd996946d-dd6dc                       0/1     ContainerCreating   0          0s
nginx-ingress-default-backend-77c7c664bb-2nsdp                 1/1     Terminating         0          92s
nginx-ingress-default-backend-77c7c664bb-mnrvj                 0/1     ContainerCreating   0          0s
---
nginx-ingress-controller-cd996946d-dd6dc                       1/1     Running   0          21s
nginx-ingress-default-backend-77c7c664bb-mnrvj                 1/1     Running   0          21s
-- VKR
Source: StackOverflow