Any ETA on the next release? We're waiting on FireLens support for ECS.

This message was deleted.

Few questions about naming resources: • Am I correct in understanding that the resource name must be unique throughout the program for any resource, not just a resource of that kind? • What can go wrong if we don't? • Do you have any recommended best practices for naming? (e.g.

<type>-<name>

?)

Hi, I’m looking to move my team from the legacy pay-per-stack model to the new unlimited stack model, but not all of my Github org users are Pulumi users. Is there a way to configure the number of users that Pulumi bills for under the unlimited stack billing model?

Is there a way to get the organization within a pulumi program? (We're using stack references, and it helps for testing things.)

Hypothetically if I wanted to install in a truly airgapped environment but still gain the benefits of pulumi.com, could I host a central pulumi.air-gapped-dns.internal?

What is the officially supported (or at least most commonly recommended) way of downgrading the Pulumi command line tool? Motivation for asking: my team has been hit by the defect addressed in https://github.com/pulumi/pulumi/pull/3798. We would like all team members to have

pulumi preview

working on their workstations. This isn't currently the case since some team members have already upgraded to 1.9.1, which is the version exhibiting the unwanted behavior, while some others are on 1.8.1 which still works for our use case.

My reading of the docs at https://www.pulumi.com/docs/get-started/install/ suggests that those needing to downgrade should

rm

the

~/.pulumi

directory, download the 1.8.1 tarball from https://www.pulumi.com/docs/get-started/install/versions/ then extract the tarball and put extracted binaries somewhere on $PATH. Is there anything else we should watch out for?

Other than experimentation, how can I tell which resources are auto-named by default? (Repost due to no response last time around.)

Hello everybody! Would you have any updates on this feature request: https://github.com/pulumi/pulumi/issues/2307? We are looking for some elegant way to load defaults from the template?

Hi. I'm attempting to create an Azure

ZipBlob

resource, but without the blob that was previously created being deleted after every run of

pulumi up

. Each blob is uniquely named. I have discovered

CustomResourceOptions

, and from those options I've used

deleteBeforeReplace: false

, assuming this would stop to resource being deleted before it was replaced. However, this is not the case, and the blob previously created by the Pulumi resource is still deleted. How can I keep blobs previously created by the Pulumi resource?

const zipBlob = new azure.storage.ZipBlob("storageBlob", {
	name: `MyApp.${version}.zip`,
    storageAccountName: storageAccount.name,
    storageContainerName: storageContainerForReleases.name,
    type: "block",
    content: new pulumi.asset.FileArchive("MyFileArchive"),
}, {
	deleteBeforeReplace: false
});

I have also tried to use the

CustomResourceOptions

of

protect

, but this caused the

pulumi up

operation to fail because Pulumi still tries to delete the resource, but can't because of the protection.

Hey guys I have a genuine question about pulumi and cicd, I currently use gitlab + pulumi and I was wondering how do you guys handle rollbacks, with your deploy. In helm, this is abit more straight forward. Have you guys come up with a good solution to this problem? Specifically with k8s. just an open discussion in a thread would be great.

Hello I was wondering how I could reference an already existing role in aws without creating one in the typescript aws module.

I tried to use the getRole data source but it seems like using its assumeRolePolicy with the role constructer creates a role in aws

Is there any way to reference terraform state as an input to pulumi? I’m looking at a model where I spin up a minimum infrastructure with terraform and then run pulumi in that infra to plug other components onto it on demand, but to do that I need to know the IDs of everything I’m hooking up to (I can export values and pass them in, but I’m wondering if there’s a more convenient way)

In the pulumi docs, I’ve seen “name”, “logical name”, and “resource_name” all used apparently interchangeably (see https://www.pulumi.com/docs/intro/concepts/programming-model/#resources , https://www.pulumi.com/docs/intro/concepts/programming-model/#urns ) which is confusing me and my team when we’re talking about these entities. Is one of these monikers preferred over the others?

Can somebody explain to me options for storing state for Pulumi. I did not find much answers in documentation. It just said either locally or on pulumi server. Terraform obviously has an option to use storage account, ARM does not require it at all. So what are the free, selfhosted options of using Pulumi? Hosting state locally obviously is not a solution since no teams are exactly 1 person.

can pulumi do automated rollbacks yet?

Anyone know if there’s any way to add a scaling policy to aws fargate services like you would through their UI?

Hi, I’m experiencing a crash with “pulumi preview” when I try to install the CRD definition for cert-manager (using Pulumi’s k8s.yaml.ConfigGroup). Did anybody see this kind of error before?

I’m trying to create an application load balancer but I get this error saying I cannot attach an application lb to multiple subnets in the same AZ. I’m using default ages for the Application load balancer class from awsx.

Hi, I’m looking to deploy the same set of resources in a number (say 25, but might grow to 100s) of AWS accounts. In CloudFormation I would use a StackSet for this, what is the best way to do this with Pulumi?

Wanted to get some feed back on some code I wrote to loop through different conditions on creating an alarm for route53. Could I have done this in a better way?

environments = {
    'HealthChecks': [health_check_1.id, health_check_2.id],
    'Names': ['Primary', 'Secondary']
}

# Metric alarm for both primary and secondary sites
for index in range(0, (len(environments))):
    cloudwatch.MetricAlarm(
        f"{environments['Names'][index]}_site_alarm",
        __opts__=ResourceOptions(
            depends_on=[sns_topic, sns_email]),
        alarm_actions=[sns_topic.arn],
        alarm_description=f"{environments['Names'][index]} alarm for website",
        name=f"{environments['Names'][index]}_site",
        comparison_operator="LessThanThreshold",
        statistic="Minimum", evaluation_periods=3,
        metric_name="HealthCheckStatus",
        namespace="AWS/Route53",
        period=60, threshold=1,
        dimensions={
            'HealthCheckId': environments['HealthChecks'][index]
        }
    )

What name would you give to the things that Pulumi uses a provider to talk to? For example: • The Kubernetes provider lets you manage resources in Kubernetes. Kubernetes is a _____. • The GCP provider lets you manage resources in GCP. GCP is a _____. “Clouds” makes sense when talking about GCP, AWS, or Linode, but not when talking about Kubernetes – and definitely not tools like Kafka or MySQL, or the Random provider. Some of the docs (and especially the Terraform docs) talks about them as if they are the providers, but also talk about the client libraries as providers, so it’s easy to get them confused. Which term would you use to describe these things?

is this known?

Is it possible for me to use the terraform generator to generate a new Pulumi provider, e.g. for Auth0? https://www.terraform.io/docs/providers/auth0/index.html

Is there a way to convert a kubernetes manifest yaml to pulumi python SDK code ?

Is there any way to cleanup pulumi history for a stack?

I need to write a referenced stack output value to a file as a string, but functions like apply(), all() etc only produce other Output values, while I need a plain string. Previously I just used getOutputSync but it hangs with newer nodejs versions. Referenced stack outputs are known before I run 'pulumi up', so it's not clear to me, why it is so hard to get the underlying values.

@bored-river-53178 I've only used the python implementation but I've found I can do anything I like inside an Output.apply() callback, including opening and writing files. Just pass the referenced output as an "apply" input and you will be called with it resolved to its final value, and then you can do the file write inside the apply callback. The reason the underlying values are not available synchronously is that they are implemented as futures so the time required to fetch them from the other stack takes place concurrently with the rest of your stack.