pulumi-community #general

general

victorious-xylophone-55816

09/15/2020, 1:31 AM

If I already have most of the configuration values needed for Pulumi stacks in

.staging.env

.production.env

files, is there anything wrong with loading them using

dotenv

or something instead of duplicating those values in my

Pulumi.staging.yaml

etc files?

victorious-xylophone-55816

09/15/2020, 1:37 AM

type StackEnvironment = "staging" | "production"
const stack = pulumi.getStack() as StackEnvironment

loadDotenv({
  path: path.join(__dirname, `../.${stack}.env`)
})

ancient-dentist-35597

09/15/2020, 7:39 AM

Hello 👋 I'm wondering if there's a way to suppress deletes on

pulumi up

. We have a pulumi stack that deploys an app service on azure, however recently I've found that on update it removes some

appSettings

that we enter manually through Azure. we had the stack working fine and the settings weren’t being deleted for about 6 months with multiple pushes a day, then suddenly it started removing the settings (within the last week or so). I've even set the specific version of pulumi in the github actions that we originally used and it's still happening. So is there any way to make sure that pulumi doesn't delete settings? Thanks! 🙏

steep-angle-29984

09/15/2020, 10:41 AM

Has anyone written a dynamic ResourceProvider for executing an ansible playbook which does more than just calling the ansible-playbook command? I'm thinking of inventory generation and advanced diff handling by not only checking input differences but also checking differences to the live system.

🧐 1

victorious-xylophone-55816

09/15/2020, 2:12 PM

Is there a trick to getting Pulumi to be idempotent? I have an existing Route53 domain, but code that tries to create and attach the domain to stuff. When running it, I get:

aws:route53:Record (certificate-record):
    error: [ERR]: Error building changeset: InvalidChangeBatch: [Tried to create resource record set [name='_d7342dc96b265c66ab8427198a513284.xxxx.com.', type='CNAME'] but it already exists]

Similarly, trying to put this

if/else

in a custom

ComponentResource

for ACM certificates, makes it attempt to delete the certificate instead of not doing anything:

const existingCertificate = pulumi.output( aws.acm.getCertificate({ domain: parentDomain }, { async: true }) )

if (existingCertificate) {
  this.arn = existingCertificate.arn
} else {
  const certificate = new aws.acm.Certificate(
  // rest of code

Is there a way to properly guarding all of your resources with "Try to look this up already and don't create it if it exists, just return the identifier"?

incalculable-dream-27508

09/15/2020, 4:51 PM

I don't know how specific to python this question is, but here goes: I'm creating several nodes belonging to two classes of machines on separate OpenStack clusters (currently two machines each on two clusters) with

REGIONS = {
    'ams': openstack.Provider("ams", cloud="cnos_ams"),
    'lati': openstack.Provider("lati", cloud="cnos_lati")
}

and then for each of the two categories of machines creating them, and putting the objects in dictionaries like:

INSTANCES_API = {
    # pylint: disable=line-too-long
    f'webadm-{i + 1}.otp.{dc}.local': openstack.compute.Instance(
        f'webadm-{i + 1}.otp.{dc}.local',
        flavor_name=FLAVOR_DB,
...
        security_groups=SECGROUPS_API,
        __opts__=pulumi.ResourceOptions(provider=REGIONS[dc]))
    for i in range(COUNT_API) for dc in REGIONS
}

and similar with

INSTANCES_LEAF

and then for convenience also create a new dictionary with all of them

INSTANCES = {**INSTANCES_API, **INSTANCES_LEAF}

Then I'm trying to get information about the machines, ideally a dictionary of

"hostname": "access_ip_v4"

to be able to use that later to generate names (since unique are required) and contents of security groups. But so far all my attempts at that have failed. My two most recent attempts are:

for srv in INSTANCES:
    exported[INSTANCES[srv].name.apply] = INSTANCES[srv].access_ip_v4.apply

and

for srv in INSTANCES:
    exported.update(
        pulumi.Output.all(INSTANCES[srv].name,
                          INSTANCES[srv].access_ip_v4).apply(
                              lambda args: json.dumps([{
                                  args[0]: args[1]
                              }])))

pulumi preview error

cold-car-67614

09/15/2020, 5:24 PM

Hello, I filed a github issue here: https://github.com/pulumi/pulumi/issues/5366 Has anyone else gotten exceptions thrown that generally lack context on where the issue could be coming from?

incalculable-dream-27508

09/15/2020, 5:25 PM

@cold-car-67614 yeah, check my message just above yours ;)

incalculable-dream-27508

09/15/2020, 5:25 PM

Ah, no, it does say. At some points in my experimentation in didn't.

astonishing-quill-88807

09/15/2020, 5:26 PM

I have definitely seen that being the case and have posted some questions in the Python channel. I ultimately ended up figuring things out, but yeah, the stack traces are often unhelpful.

cold-car-67614

09/15/2020, 5:28 PM

I've definitely seen it before, but generally I'll just revert whatever changes I had made and slowly add stuff back to figure out where it came from. In this case though all I did was upgrade Pulumi/plugins and I'm getting this exception.

cold-car-67614

09/15/2020, 5:29 PM

Going to have upgrade trauma after this 😞

loud-egg-62954

09/15/2020, 5:57 PM

Hey folks--I'm getting an API disabled error and I can't really determine the cause. I think pulumi might be pointed at the wrong GCP project

loud-egg-62954

09/15/2020, 5:57 PM

I updated the GCP project in the gcloud CLI--does it not pick it up from there?

loud-egg-62954

09/15/2020, 5:59 PM

Ah, never mind. I was looking for pulumi config set gcp:project

loud-egg-62954

09/15/2020, 5:59 PM

Awesome work folks, loving it

billowy-kangaroo-51688

09/15/2020, 6:11 PM

What could be the reason that

dynamodb.name.get()

is not being captured in my Lambdas?

many-psychiatrist-74327

09/15/2020, 6:13 PM

👋 hello! has anyone experienced very high latency when using the extension libraries (
awsx
,
eks
) in TS? I am bringing up a kubernetes cluster and a few services running in it. I first implemented in Python to learn the framework. Then I re-implemented in TS because there are better libraries. However the latency is very different: Python, 17 seconds:

$ time pulumi preview
Previewing update (dev)

View Live: (redacted)

     Type                          Name                            Plan
     pulumi:pulumi:Stack           kubernetes-dev
     (redacted)

Resources:
    - 4 to delete
    40 unchanged


real	0m17.365s
user	0m1.624s
sys	0m2.881s

TS, almost 2 minutes:

$ time pulumi preview
Previewing update (dev)

View Live: (redacted)

     Type                 Name       Plan
     pulumi:pulumi:Stack  infra-dev

Resources:
    70 unchanged


real	1m37.499s
user	0m0.998s
sys	0m1.127s

How can I find out where this latency is coming from?

billowy-kangaroo-51688

09/15/2020, 7:39 PM

My last question was a bit uselessly short 😅

new aws.lambda.CallbackFunction(
  "handler",
  {
    tags,
    memorySize: 128,
    role: lambdaRole,
    layers: [awsSdkLayer.arn],
    callback: (event: APIGatewayProxyEvent) => {
      return handlerFactory({
        tableName: db.name.get(), // db = result of `new aws.dynamodb.Table`
        // [..] more values not being captured
      })(event);
    },
  },
);

I’ve “downloaded deployment package” & the actual name of the table is nowhere to be found, I also tried to define

const tableName = db.name.get()

before the

return handlerFactory

🤷‍♂️

✅ 1

handsome-actor-1155

09/15/2020, 8:21 PM

Hi all, just a quick question: I’m using Pulumi to create and upload images to AWS ECR via a Dockerfile and it works great. I’m just curious if there’s some configuration setting that can prevent Pulumi from invoking docker, building, and trying to push the image every time I run

pulumi up

or even

pulumi preview

. This would save quite a bit of time. Or is this something I would need to configure within the Dockerfile itself? Thanks!

chilly-garage-80867

09/15/2020, 8:23 PM

General best practices question

chilly-garage-80867

09/15/2020, 8:24 PM

what should I include in a .gitignore if I am going to share a pulumi stack configuration

loud-battery-37784

09/15/2020, 8:25 PM

@chilly-garage-80867 what language?

chilly-garage-80867

09/15/2020, 8:25 PM

Python

bitter-application-91815

09/15/2020, 9:03 PM

Hey there, just here on

<https://github.com/pulumi/pulumi-aws/blob/master/sdk/go/aws/eks/nodeGroup.go#L314>

bitter-application-91815

09/15/2020, 9:03 PM

it says,

// Identifiers of EC2 Subnets to associate with the EKS Node Group. These subnets must have the following resource tag: `<http://kubernetes.io/cluster/CLUSTER_NAME|kubernetes.io/cluster/CLUSTER_NAME>` (where `CLUSTER_NAME` is replaced with the name of the EKS Cluster).`

bitter-application-91815

09/15/2020, 9:04 PM

sure but what's the key on that tag, tags on ec2::subnet is a map, not an []string

millions-judge-24978

09/15/2020, 9:15 PM

If I am creating a pulumi program in typescript, and would like to instantiate an

aws.lambda.Function

using go code, is there a way to automate the

go build

as part of the pulumi program running? I see this:

Using Pulumi’s Asset and Archive classes, we can fetch code from anywhere – in-memory, on disk, or even over the network. Pulumi will detect changes in the contents of these assets and archives so that when you run pulumi up, diffs will be detected and updated.

It seems to me though that would be missing the necessary

go build

though

fast-magician-55948

09/15/2020, 9:18 PM

How do you put several Google Cloud Functions in a Pulumi project? I've successfully created on Google Cloud Function, following the samples. Now I want to put in a second Google Cloud Function. Any samples?

fast-magician-55948

09/15/2020, 9:28 PM

(I'm using Python)