Subsequent runs of

pulumi update

result in a stacktrace-esuqe output.

Specifically, I get several of these stacktraces:

error: TypeError: Cannot read property 'environment' of undefined
        at pulumi.all.apply (/home/kenny/compute_software/demo-applications/basic-web-container/node_modules/@pulumi/service.ts:421:59)
        at Output.<anonymous> (/home/kenny/compute_software/demo-applications/basic-web-container/node_modules/@pulumi/pulumi/resource.js:236:43)
        at next (native)
        at fulfilled (/home/kenny/compute_software/demo-applications/basic-web-container/node_modules/@pulumi/pulumi/resource.js:17:58)

Although the update was not interrupted, I tried going through this https://pulumi.io/reference/troubleshooting.html#interrupted-update-recovery but it did not help.

Hi - this is my first time using Pulumi and was wondering if anyone had any boilerplate for building an API gateway - I'm trying to set up an API Gateway proxy -> Kinesis, but it seems that both the

serverless

and

cloud

abstractions don't expose the types of interfaces I need. If anyone has something basic I can experiment with - it would be much appreciated!!!

I'm really only looking for a) minimum to create the gateway b) how to create a resource c) attach to AWS service

Ah, I see the problem. I deployed this at first:

let service = new cloud.Service("basic-web", {
    containers: {
        nginx: {
            build: "./app",
            memory: 256,
            ports: [{port: 8080}]
        }
    },
    replicas: 2
});

And then forgot I changed

nginx

to

basic_web

causing that exception. How am I supposed to rename my containers?

I tried setting

containers

to

{}

but that results in an error saying that

containers

cannot be empty.

If my container exposes a http interface on port 8080, how do I set the load balancer created from

cloud.Service

to forward port 80 to 8080 in the service?

{port: 80, targetPort: 8080, external: true}

?

I tried

pulumi update

with that set as my

ports

and received:

error: TypeError: Cannot read property 'environment' of undefined
        at pulumi.all.apply (/home/kenny/compute_software/demo-applications/basic-web-container/node_modules/@pulumi/service.ts:421:59)
        at Output.<anonymous> (/home/kenny/compute_software/demo-applications/basic-web-container/node_modules/@pulumi/pulumi/resource.js:236:43)
        at next (native)
        at fulfilled (/home/kenny/compute_software/demo-applications/basic-web-container/node_modules/@pulumi/pulumi/resource.js:17:58)

@full-dress-10026 Looking at these two issues you are seeing now - sounds like you may be hitting a bug in the library - will report back once I have more details.

@creamy-potato-29402 question regarding k8s helm integration: When I specify

namespace

on a chart via pulumi it seems this only populates the

Release.Namespace

template variable, whereas when running

helm install --namespace <namespace>

tiller seems to actively add an

metadata.namespace

entry before applying the templates to the cluster. Many charts unfortunately don’t have the

Release.Namespace

variable but instead rely on the aforementioned tiller override. Would it be possible to add a similar behaviour to the pulumi helm integration? I think that would greatly decrease confusion for users. I can probably for now manually apply the desired namespace by using

transformations

but I think it would be better UX if pulumi helm behaves a little more like

helm install

instead of

helm template

. Related github issue https://github.com/pulumi/pulumi-kubernetes/issues/200

any idea what this error message means?

global: global
    error: Short version cannot contain PreRelease/Build meta data

How would I pass a custom vpc to the

cloud.Service

ctor?

I've been playing around this evening with the terraform bridge and discovered how easy it was to port the providers to pulumi packages. I've gone ahead and made an initial pass at a digitalocean provider here: https://github.com/jaxxstorm/pulumi-do is this something you guys want to support? Or should I modify to publish to my own NPM namespace?

don't think it's ready for prime time just yet but it works pretty nicely

Please choose a stack, or create a new one: digitalocean-dev
Previewing update of stack 'digitalocean-dev'
Previewing changes:

     Type                 Name                                Plan       Info
 +   pulumi:pulumi:Stack  digitalocean-test-digitalocean-dev  create
 +   └─ do:core:Droplet   test                                create

info: 2 changes previewed:
    + 2 resources to create

This is excellent! Thank you for doing this. We always love seeing new providers being brought up and we'd be happy to work with you to get it transitioned to an official

@pulumi/...

package. We've done this with OpenStack and VMWare, both of which were community contributions also 🙂

So... stuff is working better and better. What's the best/canonical way to add CRDs? I got ingress and cert-manager helm charts deployed. Now I'd like to deploy various resources defined by each.

is it possible to define some custom resource cleanup hooks or create sort of a wrapper which is able to do that? I have a k8s statefulset and would like to delete the PVCs of that statefulset after it’s being destroyed.

hey @creamy-potato-29402 can you give a quick update on whether and when https://github.com/pulumi/pulumi-docker/issues/16 and https://github.com/pulumi/pulumi-docker/issues/14 will be implemented? Those are for me currently the two biggest issues because in literally every stack I have at least one docker image (in one it’s actually 10) and those 2 issues cause very noisy and slow previews.

I had to leave for important family business this week, so I didn’t get a chance to follow up. @big-piano-35669 @white-balloon-205 I have thoughts on this, but it’s also not my baby. What do you think?

Luke does claim to be doing it for M18, so my guess is it will get done for M18. (I don’t want to put words in his mouth, but he does tend to be the most responsible of all of us in keeping his backlog accurate.)

Hmm.. is it possible to launch a fargate container with lambda through an api gateway and retrieve the cloudwatch container logs, send it back as a response to the api gateway call, all within Pulumi?

Is there a recommended approach for using CloudFormation templates with Pulumi? For example, the database we use has an automated deployment via the AWS Marketplace which uses CF templates. It'd be nice to be able to deploy my whole cluster with Pulumi though.

It looks like resource naming got more strict? I'm trying to make a new

CustomResource

and it complains about what I pass to

super

call... E.g.

class MyFunctionApp extends pulumi.CustomResource {
    constructor(name: string, options: MyAppServiceOptions) {
        super("azure:mikhail:MyFunctionApp", name);

gets

unrecognized resource type (Check): azure:mikhail:MyFunctionApp

I created a new project named

ComputeSoftware/infra

and it failed with this error:

Diagnostics:
  global: global
    info: error: It looks like the Pulumi SDK has not been installed. Have you run npm install or yarn install?
 
error: failed to load language plugin nodejs: could not read plugin [/home/kenny/.pulumi/bin/pulumi-language-nodejs] stdout: EOF

Are you allowed to name projects like that?

<org>/<project>

?

Another note: On the Pulumi app it says to delete the stack I need to run this command

pulumi destroy ComputeSoftware/infra-dev

. That command fails with the message

error: unknown command "ComputeSoftware/infra-dev" for "pulumi destroy"

.

it’s just

pulumi destroy

, right?