What we also see:

kubectl get scaledobject --all-namespaces
error: the server doesn't have a resource type "scaledobject"

It’s the same error message as the installer gets.
But: We don’t want to use autoscaling and we don’t have Keda deployed in our Kubernetes cluster.

The docs say:
If support for auto-scaling the deployments is required KEDA 2.8 must be deployed into the cluster.

Are the docs wrong? Is a KEDA deployment necessary even when not using autoscaling? (That would be realy “ugly”.)

Regards
Tobias

Hi Tobias,

I recently deployed 6.11.0.7 into K8S and this install also didn’t had KEDA installed in the cluster. I only got a warning during setup but the deployment succeded. I guess, you have Helm available on the linux machine and it’s configured in the PATH system variable? I did run the Installation from a Windows VM, so not sure if this might be a Linux related issue.

Did you provide only the changed values from default in your values.yaml? Auto-scaling is disabled by default, so you don’t have to include it in your custom values.yaml file.

Best regards
Roman

Hi Roman,
is it possible to provide the 6.11.0.7 to our ftp server and the images as well?
We’d like to check if our errors may belong to the version we try to install

Best Regards,
Steffi

Hi Steffi, Hi Tobias,

please check your mails! :)

Best regards
Roman

Thank you very much Roman!!

Hi Roman!

thanks for your answer.

We ware able to run the install successfull now.

Still getting warnings in the installer logs about KEDA CRDs (Custom Resource Definition in K8s).
It seems to be the case that we don’t need KEDA to proceed. But the warings are quite misleading if you ask me.

WARN Installer_Shared.Kubernetes.HelmDeployer b - ThreeSixty.Shared_Classes.ProcessException: error: the server doesn't have a resource type "ScaledObject" at ThreeSixty.Shared_Classes.ProccessRunner.Run(String fileName, String arguments, String input, String workingDirectory) at Ifs.Kubernetes.Kubectl.GetResources(String context, String ns, String resourceType, IReadOnlyDictionary`2 labels) at Installer_Shared.Kubernetes.HelmDeployer.b(String A_0, String A_1, String A_2)

Regards
Tobias

That’s great! :)

Yes, I agree. I also mentioned this to the Product Team. I will raise a ticket to get this fixed in the next release.

Best regards
Roman

For the sake of completness:

The reported Issue was fixed by using PSO Installation Package 6.11.0.7.

Best regards
Roman

I installed the version 6.11.0.7 today but unfortunatley we still getting the same error message:
 

WARN Installer_Shared.Kubernetes.HelmDeployer PauseDeployments - ThreeSixty.Shared_Classes.ProcessException: error: the server doesn't have a resource type "ScaledObject" at ThreeSixty.Shared_Classes.ProccessRunner.Run(String fileName, String arguments, String input, String workingDirectory) at Ifs.Kubernetes.Kubectl.GetResources(String context, String ns, String resourceType, IReadOnlyDictionary`2 labels) at Installer_Shared.Kubernetes.HelmDeployer.PauseDeployments(String context, String ns, String installName, TimeSpan waitTimeout)

Hi Steffi,

I’m a bit confused as Tobias wrote “We ware able to run the install successfull now.” - did the Installation finish or not? If the Installation finished and you only got this warning in the Installer log, than you can ignore it when you don’t use auto-scaling. I created also a Case for this: CS0134375 

Best regards
Roman

Hi Roman, it’s just this error message but the installation will proceed and is completed then. I justed wanted you to know that it does not seem to be solved in 6.11.0.7 as you mentioned in your reply before. But maybe I got it wrong.

Best Regards,
Steffi

Thanks for your reply, Steffi.

Let me explain:

The severity of this event logged is WARN. A error is shown in the Installer actually as ERROR.

I thought that maybe this fix mentioned in the release notes of 6.11.0.7 has something to do with the issue you faced in 6.11.0.5:

INST  TEPSO-3031      Fixed an issue with updating the Scheduling System Database in a Kubernetes environment

Apparently, something has changed between both versions as you were able to complete the installation process - although you had the same warning logged again. Mabye it’s how the Installer handles this warning (e.g. in 6.11.0.5 it caused the Installation to stop, while in 6.11.0.7 it actually was handled like a “warning” and the Installation finishes). However, I was expecting that you will receive the warning again as I had it as well with that version. That’s why I created the above mentioned case to request removal of this warning as it’s incorrectly shown.

@mimagb maybe you can shed some light what changed between those versions in terms of K8S Installation? 

Best regards
Roman

Hey @roklde 

The TEPSO-3031 issue is unrelated to this and was around database upgrades.

The warning is there due to the installer trying to pause the previous install in case it was done with auto-scaling enabled. This will not negatively impact the install. Installations without auto-scaling and KEDA are supported.

Once the issue you raised gets to us we can look into making the logs here less misleading.

Regards,

Michael

Thanks Michael!

It’s strange that using a different PSO version fixed the initial issue reported by the customer. Or did you do something differently on this attempt @tobias.durai @stteschn ?

Best regards
Roman

Hi Roman!

I’m not perfectly sure what fixed the error. 

• We changed the way we fill the parameters in values.yaml.
• We saw that one of the unsuccessful installation attempts created a “broken” helm application. We deleted it manually and afterwards the installation process was successful. (That would be my main guess.)

PS: When I see warnings in a log I always get (a bit) uncertain if everything is ok or not. And in this case I don’t see a reason why the installer should check for “scaledObjects” when you set “autoscaling=false”.

Regards
Tobias