How to check reason for Helm install failure

Question

I wanted to bring up zookeeper using helm install ., but it says Error: release <servicename> failed: services "zookeeper" already exists. I don't see anything if I execute helm listtoo. Before installing the service, I checked using helm list if it already exists, and it doesn't.

How to check the reason for failure?

In which namespace is it going to create resources? Check the namespace and delete the resource/clean up manually may help — Ijaz Ahmad, Oct 18 '18 at 12:04
I'm using the default namespace. I executed `kubectl delete services/zookeeper` and then tried doing helm install. But still no luck. Funny thing is, the same yaml file was working perfectly fine just yesterday. — Bitswazsky, Oct 18 '18 at 12:19
what is you kuberntes enviroment? these objects are stored in etcd — Ijaz Ahmad, Oct 18 '18 at 12:26
if you are using minikube , stop minikube , remove the kube direcotry from , and then do a fresh minikube start , eceryhting will clean up — Ijaz Ahmad, Oct 18 '18 at 12:28
othervise connect to etcd and cleanup the key. https://stackoverflow.com/questions/47807892/how-to-access-kubernetes-keys-in-etcd — Ijaz Ahmad, Oct 18 '18 at 12:28
Let us [continue this discussion in chat](https://chat.stackoverflow.com/rooms/182091/discussion-between-bitswazsky-and-ijaz-khan). — Bitswazsky, Oct 18 '18 at 12:43

score 29 · Answer 1 · answered Jan 20 '21 at 09:55

29

I think that the simplest solution is to add the --debug flag for the installation command:

helm install chart my-chart --debug

Or if you prefer:

helm  upgrade --install chart my-chart --debug

It displays all the resources which are created one by one and also related errors which occured during installation.

answered Jan 20 '21 at 09:55

Rot-man

18,045
12
118
124

1

Well... I'm installing charts using terraform so everything could be done with just one command – holms Sep 07 '22 at 06:07

score 7 · Answer 2 · edited Dec 08 '20 at 13:55

7

Do helm list --all - Helm List

Then if you have a conflicting release then probably need to delete the release again with the --purge flag

But it could possibly be that you have a Service object named zookeeper that isn't part of a helm release or that hasn't been cleaned up. You can check with kubectl get services (or add the --all-namespaces flag if it might be in a different namespace from your context). If so then you'll want to delete resources directly with kubectl delete

edited Dec 08 '20 at 13:55

Abdullah Khawer

4,461
4
29
66

answered Oct 18 '18 at 11:58

Ryan Dawson

11,832
5
38
61

While listing all releases, I could see all (except the current one) releases were marked `DELETED`. I purged that one, but didn't help. I checked the services too, didn't find any conflict. – Bitswazsky Oct 18 '18 at 12:02
Perhaps try purging even the deleted ones? – Ryan Dawson Oct 18 '18 at 12:28

Ijaz Ahmad · Accepted Answer · 2018-10-18T13:05:06.553

5

Option 01

Access the ETCD for minikube , find and clean up the key.

Detailed procedure is here

commands:

https://gist.github.com/iahmad-khan/5d32b4070b6faf3836b932a7177095ff

Option 02 ( will lose the existing stuff )

Stop minikube

Remove the kube direcotry in user home ~/.minikube

Start a fresh minikube

edited Oct 18 '18 at 13:05

answered Oct 18 '18 at 12:33

Ijaz Ahmad

11,198
9
53
73

I deleted the zookeeper service using kubectl. removed the release from helm. Stopped minikube. Deleted the .kube folder. Brought up minikube. But unfortunately still seeing the same issue. I guess I'll try out on a different machine, and see. – Bitswazsky Oct 18 '18 at 12:55
answer updated , go to etcd and see where the key exist – Ijaz Ahmad Oct 18 '18 at 13:05
The second option is gaurreented to work , you may not have cleaned up the correct direcotry , it should be in the home of user , ~/.minikube . you need to delete that and start minikube again – Ijaz Ahmad Oct 18 '18 at 13:07
Unfortunately, deleting the ~/.minikube dir messed up my minikube setup. Now it won't start anymore. – Bitswazsky Oct 19 '18 at 09:31
i have been doing it everyday , and everytime , minikune setups things from scatch – Ijaz Ahmad Oct 19 '18 at 10:27
I use this command to install minikube `curl -Lo minikube https://storage.googleapis.com/minikube/releases/v0.30.0/minikube-linux-amd64 && chmod +x minikube && cp minikube /usr/local/bin/` Is there anything that I'm missing here? – Bitswazsky Oct 19 '18 at 14:45
it is not related to minkube installation , if you can run the minikube binary then it should be ok to delete the .minikube directory and start over , it will re-create it – Ijaz Ahmad Oct 19 '18 at 14:58

Bitswazsky · Answer 4 · 2018-10-29T08:50:11.550

So, I finally got it working. Looks like there were some errors in my yaml file. But in this process I learnt how to properly cleanup a minikube instance. I executed the following commands to bring up a fresh minikube instance, in case anybody faces similar issue. This was done on CentOS 7. Thanks to @Ijaz for pointing me to the right direction.

minikube stop && minikube delete && rm -rf ~/.minikube && rm -rf ~/.kube
rm -rf /etc/kubernetes/
rm -rf /var/lib/minikube/certs/
minikube start --vm-driver=none

How to check reason for Helm install failure

4 Answers4