pulumi-community #kubernetes

Join the conversation Join Slack

kubernetes

many-psychiatrist-74327

04/15/2021, 1:44 AM

(cross-posting in case this channel is more appropriate)

handsome-state-59775

04/15/2021, 10:13 PM

any plans of bringing kubernetesx to python?

rhythmic-actor-14991

04/16/2021, 10:05 AM

* Ingress has at least one rule that does not target any Service. Field '.spec.rules[].http.paths[].backend.serviceName' may not match any active Service

straight-cartoon-24485

04/17/2021, 12:36 AM

Can we force-delete all resources that Pulumi complains already exist? I got some undeleted clusterroles and helm charts that I'd like to simply wipe. I must've deleted the older pulumi stack, so now a bunch of items are dangling. What's the most elegant way to start off a clean slate?

straight-cartoon-24485

04/18/2021, 7:50 AM

Is there a way to query for the pulumi stack name at runtime? I understand I can setup a key-value in the configuration

Pulumi.<stack-name>.yaml

file, and am wondering if there is a better way to get the

<stack-name>

to stay DRY

kind-mechanic-53546

04/19/2021, 3:10 AM

Hey all, When running on linux and doing a

new docker.Image(...

, with a

registry: {server: {azure container registry login server here}

, I'm getting the error

Error: No digest available for image {registry:tag}

This process works on windows with the exact same pulumi version and stack (but has other issues with WSL and file ownership, speed etc..) Per this file, running

docker image inspect -f {{.Id}} imageName

in a terminal gives a correct output and pushing it with

docker push {registry:tag}

runs fine outside of pulumi It looks like this was a prior issue here and here but i'm running v2.25.2 so I'm not sure where to go

straight-cartoon-24485

04/19/2021, 1:48 PM

Can a personal stack be made publicly accessible (or cloneable)? I think I stumbled onto a linkerd bug, and wanted to submit a reproducible repo for others to easily have a look at, and attach that to a public issue on github. What are some strategies to go about sharing stacks with non-pulumi users?

limited-rainbow-51650

04/20/2021, 9:51 AM

Running in circles on this: we have our own

Service

component resource. In our own abstraction, we create a NodePort based k8s

Service

resource, retrieve the actual nodeport value from the service using apply and pass it on to a GKE specific

BackendConfig

CRD. This is the code snippet:

const healthPortNumber: pulumi.Input<number> = this.service.spec.apply((spec) => {
      const healthPort = spec.ports.find((port) => {
        return port.name === 'health';
      });
      if (healthPort) {
        return healthPort.nodePort;
      } else {
        return 4001;
      }
    });

We have a stack where the ports section of the

Service

spec contains this (taken from the stack state):

{
  "name": "health",
  "nodePort": 30449,
  "port": 4001,
  "protocol": "TCP",
  "targetPort": "health"
}

But we get

undefined

as the value for

healthPort.nodePort

. How is that possible??

prehistoric-kite-30979

04/21/2021, 4:18 PM

~~Are there any kustomize gotchas? I'm passing an absolute path directory in and getting nil pointer panics in go.~~ ~~Update: it doesnt appear to be an issue finding the directory because if I intentionally set a base in the kustomization.yaml that doesn't exists it gives me the expected error.~~ Second update: it looks like the issue is that my customize was using the built in LabelTransfomer

prehistoric-kite-30979

04/21/2021, 6:05 PM

^^ So I believe the issue is that built in transformers don't work in Pulumi kustomize but do in standard kustomize

colossal-plastic-46140

04/22/2021, 1:53 AM

Hi Folks, I was wondering if anyone had advice on how to structure a number of component resources that share a single namespace. Each of the component resources inherit their configs from a base object which includes the namespace. Keep in mind that I am using python. Any advice is appreciated.

miniature-leather-70472

04/23/2021, 3:53 PM

Has anyone worked with the Prometheus helm chart with Pulumi? I've been able to deploy it initially without issue, however when it comes to upgrading the version it fails, it seems that Pulumi creates a new replica set and waits for it to be ready before turning the old one down to 0, however Prometheus server pod can't start becuause it can't get a lock on the volume, because the old pod is still using it. Going in and setting the old replica set to 0 resolves it. Anyone have an idea on resolving this?

billowy-army-68599

04/23/2021, 4:53 PM

hi community! would love to get some thoughts about Pulumi packages here: https://twitter.com/briggsl/status/1385632978504613890?s=20

colossal-twilight-92354

04/23/2021, 6:34 PM

hi everyone - new to pulumi, and curious whether this is a common experience for people: I am trying to manage some existing deployed charts, and am trying to import them via

pulumi import <kubernetes:helm.sh/v3:Chart> mychartname mychartname

, but keep getting

error: Preview failed: resource 'mychartname' does not exist

(despite being able to successfully import k8s resources independent from importing the chart itssself). How do others handle importing chart state?

proud-pizza-80589

04/24/2021, 8:06 AM

Deploying helm charts with secrets always recreate the secret, even if the data is not changed. Is there any way to prevent it from doing so? I used “protect: true” on such a chart and lost all update capabilities 🙂

sticky-hydrogen-53391

04/24/2021, 8:22 AM

Hi team, am I missing something or is there no way to pass in helm values in a yaml file, like you would with the helm cli? Ie. Using a remote chart but passing in my own values.yaml via '--values' parameters. It seems convoluted and difficult to manage to have to turn my simple(ish) yaml file into a c# dictionary to override the default values of the chart...

proud-pizza-80589

04/26/2021, 3:48 PM

Do you need Helm installed to use Helmcharts via Pulumi?

purple-plumber-90981

04/28/2021, 6:58 AM

when im creating a namespace with pulumi, what is the right structure to not get the random 8 character string on the end per

<https://www.pulumi.com/blog/infrastructure-as-code-resource-naming/>

my_namespace = k8s.core.v1.Namespace("mynamespace")

gets me a namespace called mynamespace-778h547k when i really just want “mynamespace”

kind-mechanic-53546

04/28/2021, 7:00 AM

That would be

... { metadata: { name: yournamehere,...

☝️ 1

purple-plumber-90981

04/28/2021, 7:02 AM

and in an aws eks cluster creation like :

eks_cluster = aws.eks.Cluster("my-eks-cluster", opts=eks_opts, **eks_cluster_config)

how would i achieve a cluster called “my-eks-cluster” rather than “my-eks-cluster-3ef8ab7” ?

kind-mechanic-53546

04/28/2021, 7:04 AM

most resources have an explicit name property somewhere in the args which will override the naming convention

kind-mechanic-53546

04/28/2021, 7:05 AM

Seee https://www.pulumi.com/docs/reference/pkg/aws/eks/cluster/#inputs

icy-jordan-58549

04/28/2021, 3:04 PM

CustomResourceDefinition "<http://kafkas.kafka.strimzi.io|kafkas.kafka.strimzi.io>" is invalid: metadata.annotations: Too long: must have at most 262144 bytes

purple-plumber-90981

04/29/2021, 6:33 AM

in kubernetes storage modules, is v1 or v1beta1 more recent ?

many-psychiatrist-74327

04/30/2021, 9:56 PM

cross-posting since it involves

k8s.yaml.ConfigFile

👀 1

purple-plumber-90981

05/03/2021, 9:08 PM

hi folks, I want to use a single pulumi stack to build an eks cluster and create a namespace in it. In pulumi/python, is there a way (in the same stack) upon building an eks cluster with pulumi, to then have that new cluster config used by all subsequent pulumi_kubernetes operations ? ie without needing to

aws eks --region us-east-1 update-kubeconfig

and

pulumi config set kubernetes:context

outside of the stack

colossal-australia-65039

05/03/2021, 9:35 PM

pulumi preview --diff

nor

pulumi up

shows the detailed diff for my deployments. For example, when my actual diff is an image change all I see is:

~ kubernetes:apps/v1:Deployment: (update)
      ~ spec: {
          ~ template: {
              ~ spec: {
                  ~ containers: [
                      ~ [0]: {
                            }
                    ]
                }
            }
        }

I'm on version 3.1.0 for pulumi, and my npm libs are

"@pulumi/kubernetes": "^3.0.0",
        "@pulumi/pulumi": "^3.0.0",

this happened before the 3.0 upgrade but i only decided to ask about it now. Anyone know what's going on and how to fix it?

lemon-monkey-228

05/04/2021, 2:30 PM

If a resource in Kubernetes is altered/removed/whatever, how can I reconcile it with the Pulumi stack/

lemon-monkey-228

05/04/2021, 2:31 PM

Reapplying the stack means nothing happens because it doesn’t detect a state change

lemon-monkey-228

05/04/2021, 2:31 PM

pulumi stack export

then removing the resource into a

pulumi stack import

caused the

create

message to appear on the stack preview, but it didn’t actually happen

Title

lemon-monkey-228

05/04/2021, 2:31 PM

pulumi stack export

then removing the resource into a

pulumi stack import

caused the

create

message to appear on the stack preview, but it didn’t actually happen

bumpy-summer-9075

05/04/2021, 2:33 PM

Have you tried

pulumi refresh

lemon-monkey-228

05/04/2021, 3:01 PM

ah, didn’t know that was there!

Will give it a go next time, thanks

😃 1

purple-plumber-90981

05/05/2021, 2:30 AM

you can also manually remove a single resource like

pulumi destroy --target urn:pulumi:<thing>

and it should remove from state

lemon-monkey-228

05/05/2021, 8:33 AM

👍

View count: 5