Hi Guys. Is it possible when we provisioning for example an s3 bucket, passing the credentials by arguments instead of having it on a route like ~/.aws/credentials file? I mean, using this example: https://github.com/EvanBoyle/automation-api-examples/blob/main/inline_program/main.go It read credentials from ~/.aws/credentials or from env-vars, My idea is for example each request to an API that receives those credentials by arguments.

How should I set up structured configuration via the automation API? I have the following snippet in my code:

stack.SetConfig(ctx, "osimis.orthanc:config.name", auto.ConfigValue{Value: customer})
	stack.SetConfig(ctx, "osimis.orthanc:config.domain", auto.ConfigValue{Value: fmt.Sprintf("<http://%s.orthanc.osimis.io|%s.orthanc.osimis.io>", customer)})
	stack.SetConfig(ctx, "osimis.orthanc:config.version", auto.ConfigValue{Value: version})

In my stack, I retrieve the settings via:

const demoConfig = new pulumi.Config('osimis.orthanc').requireObject<orthanc.OrthancServiceConfig>('config')

But when running my automation program, I still have the error:

error: Missing required configuration variable 'osimis.orthanc:config'

https://www.pulumi.com/docs/intro/concepts/config/#structured-configuration

In short, I need to find the programmatic version of this CLI command:

pulumi config set --path osimis.orthanc:config.name demo

Hmmm, looking at the source, it seems it ain’t supported. No reference to

--path

at all to be found: https://github.com/pulumi/pulumi/blob/3be1ea9592b305df8aeb83c3810aee9bb8adeb7a/sdk/go/x/auto/local_workspace.go#L166-L184

I have a container running on K8s with an automation API application. The application uses

npm install

in a

Setup

func on the

GitRepo

setup like adviced here: https://pulumi-community.slack.com/archives/C019YSXN04B/p1600959484002700?thread_ts=1600959097.002500&cid=C019YSXN04B The app is running using a sub-image of

pulumi/pulumi-nodejs:2.11.2

. But when executing this, the

npm install

bails out. When trying this manually via an attached terminal, the

npm install

seems to hang around 75%. Package being installed at that moment was

grcp-js

, but not sure if it is related to that. Eventually, the process is killed. sigh

Initial NodeJS support is out for code review. We should have something ready for interested parties to kick the tires with next week's CLI release. https://github.com/pulumi/pulumi/pull/5347

@salmon-account-74572 has a pretty awesome talk going on right now that demonstrates a combination of Automation API, kustomize, and Cluster API: https://cloudengineering.heysummit.com/talks/exploring-the-intersection-of-clusterapi-and-infrastructure-as-code/

Thanks 😄

@salmon-account-74572 I loved your talk, thanks so much!

@brave-caravan-6336 gave a great talk on deploying multi-region Kubernetes, including a cool demo using Automation API to automate the process https://cloudengineering.heysummit.com/talks/shipping-a-new-multi-cloud-multi-region-kubernetes-platform-at-snowflake/

Happy to announce that the Automation API PR has been merged for NodeJS: https://github.com/pulumi/pulumi/pull/5347 We'll do a CLI release next week that will include this in the official SDK, but here's an sneak peak for those who are interested: https://github.com/EvanBoyle/automation-api-examples/tree/main/typescript/inlineProgram That example includes VSCode debug config. Here's an example of debugging an

.apply

call:

wohooo! looking forward to trying it out! 🙂

btw, one question, does the automation API still require the user to download/install the Pulumi CLI first, or is this handled automatically (not-needed?) somehow? I can see this

await stack.workspace.installPlugin("aws", "v3.6.1");

, this looks promising already.

Hi, I am a senior cloud developer at Ubisoft. We have recently decided to use Pulumi, notably the kubernetes operator, for our k8s provisioning engine we are currently building. As such, we would like to contribute to the project. Is there a proper channel of communication to help guide our potential contributions and make sure we working with the community?

The Pulumi CLI v2.12.0 is here, and with it comes the Automation API for Node.js! There's a blog post coming soon with more examples, but here some pointers to get you started: • Automation API examples have been added for node: https://github.com/pulumi/automation-api-examples • Node.js docs: https://www.pulumi.com/docs/reference/pkg/nodejs/pulumi/pulumi/x/automation/ • CLI release page where you can find v2.12.0 (required for Automation API + Node.js): https://www.pulumi.com/docs/get-started/install/versions/ • A list of known issues you can chime in on/add to: https://github.com/pulumi/pulumi/issues?q=is%3Aopen+is%3Aissue+label%3Aarea%2Fautomation-api+label%3Alanguage%2Fjavascript

The blog post that I mentioned is up. There are some really cool demos in there. I'm excited to hear what you all think! pulumi.com/blog/automation-api https://pulumi-community.slack.com/archives/CB36DSVSA/p1603124967060500

How high is the support of the other storage backends on you list?

I’d really like to build a POC around our current infrastructure to provision everything we need for a new customer. However we currently have all of our state files in Azure storage. Having access to these backends would be A+ for my use case.

Pulumi newbie here - just read about automation API. Is my understanding correct when I say that - earlier pulumi gave a cli with an ability to define infra in your fav prog lang - so the interface to the user was the

pulumi

command and now with automation api -

pulumi

command itself is a lib call and can be wrapped over custom REST APIs to abstract the infra management process?

Is pulumi automation api same as https://github.com/hashicorp/terraform-cdk - just trying to understand the differences. In case of TF CDK it says -

The CDK for Terraform project helps users define infrastructure resources using supported programming languages and generates a Terraform configuration in JSON

But in case of Pulumi - there is no interim DSL generated right? So the infra deployment capability is provided by the lib as a feature and does not need an extra step of running the deployer i.e. to do a

pulumi up

or

terraform apply

Thanks for the detailed explanation @lemon-agent-27707 - this helps clear my doubts.

We’ve scoped the effort for Python and C# including details on timeframe. Please take a look at the issues if you’re interested in following process or contributing to help accelerate the timeline. • Python • C# I'll be out on vacation this next week, but members of the team will be around for support if there is anyone who wants to get started contributing!

Is there a way to set array config items in the new node SDK? Or is it just basic string values at the moment

maybe a silly question, but how do I keep the console logs great when I transition to the automation api? the cli version (left) is much more concise than the automation api one (right). Is there any workaround for keeping the great experience I am used to having?

Is it possible to use dynamic resources (i.e.

extends pulumi.dynamic.Resource

) in a nodejs inline program using the Automation API? I get an exception:

Error: Cannot find module '@pulumi/pulumi/cmd/dynamic-provider'

as well as

error: could not read plugin [C:\Users\barcl\.pulumi\bin\pulumi-resource-pulumi-nodejs.cmd] stdout: EOF

. The installation works perfectly well using the traditional CLI-driven method

If I call

refresh()

on a stack that doesn't exist, what happens? Is that an error or no-op? (Using TypeScript if it matters)

If I don't want to check in the stack config files, can I simply getOrCreateStack, set all the config again then run

up

?

I'm trying to run a program using the automation API and getting this error:

Error: Program run without the Pulumi engine available; re-run using the `pulumi` CLI
at requireTestModeEnabled (C:\Users\Justin\Source\foo\internal-docs\deploy\node_modules\@pulumi\pulumi\runtime\settings.js:108:15)
at Object.getMonitor (C:\Users\Justin\Source\foo\internal-docs\deploy\node_modules\@pulumi\pulumi\runtime\settings.js:190:13)
at Object.readResource (C:\Users\Justin\Source\foo\internal-docs\deploy\node_modules\@pulumi\pulumi\runtime\resource.js:48:32)
at new Resource (C:\Users\Justin\Source\foo\internal-docs\deploy\node_modules\@pulumi\pulumi\resource.js:204:24)
at new CustomResource (C:\Users\Justin\Source\foo\internal-docs\deploy\node_modules\@pulumi\pulumi\resource.js:303:9)
at new StackReference (C:\Users\Justin\Source\foo\internal-docs\deploy\node_modules\@pulumi\pulumi\stackReference.js:44:9)
at C:\Users\Justin\Source\foo\internal-docs\deploy\index.ts:64:193
at Generator.next (<anonymous>)
at C:\Users\Justin\Source\foo\internal-docs\deploy\index.ts:8:71
at new Promise (<anonymous>)

I'm sure i'm missing something obvious. (thread)

Can I set tags on a stack using the API, and if so how? Link to example or docs is fine ...

I'm creating a cloudfront distribution with attached lambda functions ("Lambda@Edge"). The ARN and Version properties on my lambda functions (

aws.lambda.CallbackFunction

) are resolving to

undefined

inside

apply

. Why would that be? (thread)