anybody else having issues with modifying/deleting a launch config from EKS nodes?

consistently get:

error deleting Autoscaling Launch Configuration (monitoring-nodeLaunchConfiguration-1dc3c51): ResourceInUse: Cannot delete launch configuration monitoring-nodeLaunchConfiguration-1dc3c51 because it is attached to AutoScalingGroup monitoring-475ed714-NodeGroup-JKFQB8KKR78N

which means you can’t destroy a cluster

Just wondering if there’s anything we can do to push this issue forwards? https://forums.aws.amazon.com/thread.jspa?threadID=307382

hi all - does anyone have any info on how terraform versions work with Pulumi’s

terraform.state.RemoteStateReference

? I’m just trying it out for the first time, but get a version mismatch:

terraform:state:RemoteStateReference (synchronicity):
    error: Preview failed: error refreshing Terraform state: state snapshot was created by Terraform v0.12.6, which is newer than current v0.12.0; upgrade to Terraform v0.12.6 or greater to work with this state

My installed version of terraform is 0.12.6, so I assume Pulumi is bundling 0.12.0?

Hey folks, is there a way I can set the

policy

on the

RestApi

object when using the

awsx/apigateway.API

approach? I can't see a way of configuring the details beyond what is available on the API object, am I gonna have to use the

aws/apigateway

bits to build this up manually?

Hey guys, with GKE I am setting a cluster stack separately. In the app stack I try to reference using StackReference, on the pulumi platform, both the k8sProvider and kubeconfig are showing as outputs. But if I try to reference these in my app deployment, it does not get located and defaults to kubectls current context. I reference the stack with:

export const clusterStackName = config.require("clusterStackName");
const clusterStack = new pulumi.StackReference(clusterStackName);

The stack name uses <organisation>/<project>/<stack> so: ianbarry/infrastructure/ci as an example. I reference this stack like follows in both service and deployment CustomResourceOptions object:

provider : new k8s.Provider(clusterStackName,{
    kubeconfig: clusterStack.getOutput("kubeconfig")
})

Anyone know why a AWS policy fails to be deployed when only the description changes?

Hi guys, I have a question. I have two projects, with S3 backend. Project A with stack default and Project B with stack dev. Now I want to get the stack output of Project A in Project B. How can I achieve this? Thanks in advance.

Had a question on install - what is a good way to run curl -fsSL https://get.pulumi.com | sh to install pulumi into a specific directory

Like with sdkman you can specifically set the SDKMAN_HOME to then run a similar sdkman install into that dir. Is there something similar for pulumi?

I know I could download and unpack the tar.gz file into a directory but I would need to know the version number every time as part of a bash script - thus would be very helpful to be able to run the setup script to install into /opt where I keep my tools

Thanks for your help

Any idea why a nodepool is trying to be replaced and the only thing the diff shows different is the provider?

is the @pulumi/azure 19.4 broken? it seems to rely on @azure/ms-rest-js and its throwing all sorts of errors

Hi, has anyone tried to run pulumi in docker:dind? I tried to run it but it seems like it does not recognize the pulumi executable

so, you're using the pulumi image from inside a container running the

docker:dind

image?

(the topology of what you're doing is unclear)

pulumi preview/up

just exits with code 255 and no error, not preview what do I do wrong? index.ts:

import * as pulumi from "@pulumi/pulumi";
import * as gcp from "@pulumi/gcp";

const config = new pulumi.Config("gcp-bootstrap");

// Create a GCP resource (Storage Bucket)
const gke = new gcp.organisations.Project("gke01", {
    orgId: config.require("organisation")
});

// Export the DNS name of the bucket
export const gkeId = gke.id;

tail of

pulumi preview

Logging in using access token from PULUMI_ACCESS_TOKEN
Previewing update ($CORP/prod):

Permalink: <https://app.pulumi.com/$CORP/gcp-bootstrap/prod/previews/242b9dd0-b832-4f99-94fd-de87398528ca>

Was running no problem earlier. Then ran

pulumi preview

, froze the console, broke out and this started to happen. A reboot didn't fix it.

This will probably seem an odd question, but is there any way to use the pulumi API’s without the declarative plan and apply bits? For example, I’d like to just have a program that runs

const deployment = new k8s.apps.v1.Deployment(...)

and have it create it right then and there.

Does pulumi/gke drain the nodes when doing a replacement on the node pool in a way that a PDB would apply? Because when pulumi goes to replace a pool, my pods go down even though I have PDBs.

having exactly same probelm as in https://github.com/pulumi/pulumi/issues/2651 but with GCP provider 😕

I am pretty sure problem is in custom CA certificates we have to use and that error is not properly propagated 😞

does anyone know how kubernetes namespaces work with the

getResourceProperty()

method? All the examples I’ve seen look like this:

guestbook.getResourceProperty("v1/Service", "frontend", "status")

but I get an exception unless I prefix the resource name with the namespace:

mysql.getResourceProperty("v1/Service", "idm-mysql", "spec")
...
TypeError: Cannot read property 'spec' of undefined

whereas this works:

mysql.getResourceProperty("v1/Service", "pulimi-test/idm-mysql", "spec")

what am I missing?

the pulumi kubernetes provider IDs are a normal kubernetes reference so yeah namespace/object

if you dont specify the namespace it's assumed to be 'default'

ah ok - so in examples like this everything is bound to the

default

ns? https://github.com/pulumi/kubernetes-the-prod-way/blob/master/services/wordpress/index.ts

yeah exactly