pulumi-community #general

swift-postman-59572

09/24/2019, 5:35 PM

I want to discover the IPs of some EC2 that are stood up through an aws autoscaling group. These IPS need to be included in our application environment. The standing up of the EC2s and the creation of the application environment config are done in the same pulumi project. Anyone know of a good solution?

broad-dog-22463

09/24/2019, 5:36 PM

You could shell out from your code to the AWS CLI for Lookup

👍 1

cold-car-23440

09/24/2019, 8:29 PM

does pulumi have the concept of provisioners like terraform does? Example use-case 1: Running an ansible playbook against newly provisioned infrastructure. Example use-case 2: Bootstrap infrastructure with Puppet agent

worried-city-86458

09/25/2019, 5:40 AM

Copy code

const monitoringNamespace = new k8s.core.v1.Namespace("monitoring", { metadata: { name: "monitoring" } }, { provider : provider });

const prometheusOperatorChart = new k8s.helm.v2.Chart("po", {
    repo: "stable",
    chart: "prometheus-operator",
    namespace: monitoringNamespace.metadata.name,
    values: {
        alertmanager: { persistentVolume: { storageClass: "gp2" } },
        server: { persistentVolume: { storageClass: "gp2" } }
    }
}, { provider: provider });

worried-city-86458

09/25/2019, 5:41 AM

Gives me a warning:

Copy code

Diagnostics:
  kubernetes:helm.sh:Chart (po):
    [Can't preview] all chart values must be known ahead of time to generate an accurate preview.

worried-city-86458

09/25/2019, 5:41 AM

Not sure how to make sure the namespace is created first and not get this warning

worried-city-86458

09/25/2019, 5:42 AM

Adding

dependsOn

makes no difference

rhythmic-finland-36256

09/25/2019, 8:44 AM

Implementing dynamic resource provider for several resources and sharing configuration? Any best practices for that?

thankful-optician-22583

09/25/2019, 2:06 PM

(Kubernetes) is there a best practice for creating a hierarchical config structure for the application. For example a base configurations that has app name, memory, cpu, & then consume the Pulumi.prod.yaml. Seems like pulumi is able to understand the stack and look to a specific file. Can there be a tree of config files that it consumes.

Copy code

Pulumi.base.yaml -> Pulumi.prod.yaml  &&  Pulumi.base.yaml -> Pulumi.stage.yaml

thankful-optician-22583

09/25/2019, 3:54 PM

Right now, I wrote a script to set values before running

Copy code

Pulumi config set X, Pulumi set config Y,  Pulumi up

For example, the image tag to pull from ecr, memory size. So its not sitting in a config file but its a set of values in a file that is consumed and set prior to deployment. We did this to eliminate config duplication.

nice-airport-15607

09/25/2019, 6:25 PM

https://pulumi-community.slack.com/archives/CB36DSVSA/p1569431669010400

handsome-actor-1155

09/25/2019, 6:35 PM

Hello, I have a question regarding deploying helm charts with Pulumi. I understand that tiller does not need to be installed as Pulumi expands the chart and submits it to k8s. That being said, I'm trying to deploy an operator and it's components to k8s and can do so manually with helm + tiller but if I try to use Pulumi, the second component fails to deploy. (Rest in comments)

purple-appointment-43233

09/25/2019, 7:30 PM

Hello all, I’m trying to setup an ALB that invokes a lambda function; do y’all have any examples of setting up the certificate/dns (route 53)?

thankful-optician-22583

09/26/2019, 12:56 AM

Hey guys, so I was looking at pulumi/pulumi/runtime/config.js and I noticed that the config file that we create for each stack, is set into an env variable called

Copy code

PULUMI_CONFIG

This means that when we do

Copy code

Pulumi Up

the file is loaded as a string and parsed into a map. This would mean that if we wanted to have a hierarchical config structure for the config, we need to append the base config to this env_variable

Copy code

PULUMI_CONFIG

. My suggestion seems so hacky. As a team we decided to create a wrapper class for the config that reads from the pulumi Config class and also from a file. This way we can have a one level hierarchical structure.

thankful-optician-22583

09/26/2019, 12:57 AM

I was wondering if anybody had any opinion or correction of my understanding

worried-city-86458

09/26/2019, 1:03 AM

I'd also like to have hierarchical config. There's an issue for it: https://github.com/pulumi/pulumi/issues/2307

big-caravan-87850

09/26/2019, 1:10 AM

Someone else in the issue you mentioned above would like a solution similar to the following: https://github.com/pulumi/examples/blob/master/aws-ts-static-website/Pulumi.yaml I think putting the shared configs in the Pulumi.yaml file seems like a reasonable place. If there are stack specific configs they would take precedence and it has the added benefit of not having to change any existing pulumi programs.

freezing-artist-51725

09/26/2019, 3:33 AM

I am trying to add a node pool to a gke cluster. I have the following

Copy code

gpu_node_pool = NodePool(
    resource_name="gpu-pool-0", 
    cluster=k8s_cluster.name,
    initial_node_count=0,
    autoscaling={"minNodeCount":0, "maxNodeCount":1},
    node_config={"guest_accelerators": ["count", "1", "type", "nvidia-tesla-p100"], 
                "disk_size_gb": 200, 
                "diskType": "pd-standard",
                "imageType": "COS"})

freezing-artist-51725

09/26/2019, 3:34 AM

However, I am getting this error.

freezing-artist-51725

09/26/2019, 3:34 AM

Copy code

error: gcp:container/nodePool:NodePool resource 'gpu-pool-0' has a problem: node_config.0.guest_accelerator.0: expected object, got string
    error: gcp:container/nodePool:NodePool resource 'gpu-pool-0' has a problem: node_config.0.guest_accelerator.1: expected object, got string
    error: gcp:container/nodePool:NodePool resource 'gpu-pool-0' has a problem: node_config.0.guest_accelerator.2: expected object, got string
    error: gcp:container/nodePool:NodePool resource 'gpu-pool-0' has a problem: node_config.0.guest_accelerator.3: expected object, got string

freezing-artist-51725

09/26/2019, 3:35 AM

The node config def is the following for "guest_acclerators":

Copy code

* `guest_accelerators` (`pulumi.Input[list]`)
        
            * `count` (`pulumi.Input[float]`)
            * `type` (`pulumi.Input[str]`)

busy-air-16359

09/26/2019, 10:38 AM

I’m using pulumi to manage gke cluster but everytime I’m doing a

pulumi up

it replaces all my nodepools (although nothing changed). The diff says

~nodeConfig

. Any way to find out what the diff is?

icy-engineer-10830

09/26/2019, 2:13 PM

In the terraform gcp code for an instance template there's the ability to set lifecycle data but I do not see the equivalent in pulumi. am I missing something or is the not present?

broad-dog-22463

09/26/2019, 2:18 PM

@icy-engineer-10830 can you link me to it?

cold-car-23440

09/26/2019, 2:19 PM

are you referring to the option that allows you to specify whether a replacement resource is created before or after a destroy @icy-engineer-10830?

cold-car-23440

09/26/2019, 2:20 PM

@broad-dog-22463 i think he’s referring to this https://www.terraform.io/docs/configuration/resources.html#lifecycle-lifecycle-customizations

important-librarian-11689

09/26/2019, 2:29 PM

hi there o/ I see pulumi scripts need to be invoked using a custom runtime, "pulumi". I would be interested in using pulumi as a library, from a custom k8s operator. Is this possible somehow - for example if I run the engine in a separate k8s pod?

icy-engineer-10830

09/26/2019, 2:57 PM

this is given as an example for a GCP `InstanceTemplate`:

Copy code

* const instanceTemplate = new gcp.compute.InstanceTemplate("instanceTemplate", {
 *     // boot disk
 *     disks: [{}],
 *     machineType: "n1-standard-1",
 *     namePrefix: "instance-template-",
 *     // networking
 *     networkInterfaces: [{}],
 *     region: "us-central1",
 * });

but fails on creation with:

Copy code

Diagnostics:
  gcp:compute:InstanceTemplate (instanceTemplate):
    error: gcp:compute/instanceTemplate:InstanceTemplate resource 'instanceTemplate' has a problem: "name": conflicts with name_prefix

stale-park-1622

09/26/2019, 4:26 PM

Hi, When I build a docker image for ECR using the awsx package for node i.e.

Copy code

awsx.ecr.Repository

Copy code

repository.buildAndPushImage

how can I add a custom tag to the image? right now, it is being tagged as 'latest'. I would like to add a custom tag name for the image. Is this possible? thanks

nice-cat-91582

09/26/2019, 4:44 PM

Hey folks - I'm getting a rather confusing error when trying to create a lambda via a

lambda.CallbackFunction