Hi, I'm using Pulumi with AWS and I always keep everything (plugins, libraries) up to date. But I realise that Pulumi is downloading both 2.x and 3.x plugins. Even if I delete the 2.x plugin it always gets downloaded. Considering that I have the 3.x plugin already there, why would this happen? I'm just trying to understand this behaviour.

I seem to have gotten into a bad state after doing a

pulumi up

, and I don’t know how to get out of it… 1. My k8s deployments were running fine. 2. Run

pulumi up

with a new container image for my deployment. Pods get into crash loop and deployment fails. 3. Subsequent

pulumi up

commands always fail with

error: transport is closing

. Not very informative error… I’ve tried

pulumi refresh

and also running

up

with

--logtostderr -v=7

, but I don’t see anything useful

Is it a know issue that init a stack with azure key vault takes a long time or hangs? Or am I doing something wrong? Using a command like:

pulumi stack init --secrets-provider="<azurekeyvault://mykeyvaul.vault.azure.net/keys/mykeyname>" Dev

I get the message that the stack was created, but then it stops for some reason.

Looking for any docs, samples, etc. on how to implement a CustomResource. I read the Programming Model, but I was hoping for something lower level (i.e. code).

Hi all, is there a way to specify the latest version of the plugin to use? I've updated python and it's still trying to use an old version.

Diagnostics:
  pulumi:providers:vsphere (default_2_8_0):
    error: no resource plugin 'vsphere-v2.8.0' found in the workspace or on your $PATH, install the plugin using `pulumi plugin install resource vsphere v2.8.0`

How can I force it to use

2.10.1

pulumi plugin ls
NAME     KIND      VERSION  SIZE   INSTALLED    LAST USED
vsphere  resource  2.10.1   63 MB  6 hours ago  6 hours ago

TOTAL plugin cache size: 63 MB

I had a protected resource and now I have set

protect=False

but it still refuses to delete the protected resource.

Hey everyone, do you know of an example of using Pulumi to build a local Docker image and deploy it to a local K8s? No cloud involve.

This might sound like a stupid question, but not sure. Let's say we are using azure blob storage to store the state so we use that to

--cloud-url

to log in, and I have also set all the azure config with

pulumi config set azure:...

. When we then execute pulumi, what credentials are used? Is it the credentials that I set using

pulumi config set azure:..

or is it some other credentials? For me it sort of make sense it it where two different set of credentials, since one is used to authenticate against the pulumi state and keyvault one might use, but the config you set with

pulumi config set azure:...

is the credentials you use for actual provisioning of resources.

I was on the dotnet documentation for AWS Glue and noticed that in the code examples that the character sets and entity sets are not properly parsed out

using Pulumi;
using Aws = Pulumi.Aws;

class MyStack : Stack
{
public MyStack()
{
    var awsGlueCatalogDatabase = new Aws.Glue.CatalogDatabase("awsGlueCatalogDatabase", new Aws.Glue.CatalogDatabaseArgs
    {
        Name = "MyCatalogDatabase",
    });
}

}

hi all, i’m new to pulumi and i was wondering is it possible to keep the state under version control, or is it a bad idea?

I do that on some projects where the IaC is not run often, sensitive contents should be encrypted with the password

sure, in private repository, we did something similar with terraform back in the days, hence i was thinking doing this with pulumi as well, maybe. thanks for your response.

I would have used the S3 backend but its incredibly slow.

Speaking of S3, there's apparently some real bugginess with the handling of BucketObject's. Trying to sync it with a local folder has unexpected side effects like objects not being overwritten, objects not being deleted - I seem to have sorta hacked a fix together by doing pulumi up with the assets ignored before the real pulumi up, but apparently some assets are still not deleted.

It seems like NodePools don't de-provision properly. Most resources seem to accept a 404 from the API as indicating a resource has been deleted successfully, NodePools on GKE don't

Has anyone run into a

Duplicate resource URN

for their stack? For context, I am running

--refresh

in my

pulumi up

CI workflow.

<PROJECT-NAME>-<STACK-NAME> (pulumi:pulumi:Stack)
error: Duplicate resource URN 'urn:pulumi:<STACK-NAME>::<PROJECT-NAME>::pulumi:pulumi:Stack::<PROJECT-NAME>-<STACK-NAME>'; try giving it a unique name

Is there a way to connect to Pulumi from Gatling Load Tests?

For a Pulumi employee: is it possible to enable issue tracking on the Random provider, please? https://github.com/pulumi/terraform-provider-random

how do I convert a string to a input<string?

Hey everyone. Is there an example or type of setup when one can use Pulumi Kubernetes provider like Skaffold in which it automatically detects file changes, rebuilds and redeploys containers?

Hi, how do I convert an Output to a float? Thanks

Hi all, i need to chain provisioning of an EKS cluster with resource being added to that cluster. My initial approach would be to dynamically create the k8s provider with outputs from the EKS cluster, but that at least has the impact of not showing the k8s resources in the preview. The k8s resources start showing up after the cluster has provisioned and pulumi can resolve the chain of dependencies. Is this approach kosher? Is there a better way to go about it?

Resource exclusion / partial run in Pulumi: I'm trying to exclude a set of resources from being considered for update - sort of a "partial run". If I exclude them from the code Pulumi will try to remove them 😉 any idea how to simply 'exclude' them temporarily?

Hi, how would I resolve a

pulumi.Output[List]

early (or similar)? I importing this list from another stack and wish to iterate through the outputs to create resources.

pulumi.Output#apply

suggests I should not be creating resources using

func

any ideas what’s this?

error: resource mm/list-query-engine was not successfully created by the Kubernetes API server : failed to determine if the following GVK is namespaced: /v1, Kind=List

Is there a way to to achieve the inverse of

pulumi up -t "urn"

? While waiting for a fix for https://github.com/pulumi/pulumi-aws/issues/1167 it would be an easy work around if I could run pulumi up for everything except the elasticache target. It is a fairly large stack so suppling all the targets except elasticache to the -t argument isn't a feasible option

hi all. I would like to ignore, at provider level, some tags. I’m doing somethink like:

but with no luck