Hi, I was wondering if there was a way to force deployment of a new docker image in an

ecs.EC2Service

?

Has anyone built a provider to manipulate resources within Codefresh? I'm not talking about using Pulumi from inside Codefresh, but creating pipelines, triggers, etc. using Pulumi.

I have a resource that is always tainted (spurious diff) - how do I go about tracking down if the issue is with pulumi or terraform plugin? Is state diff something that pulumi handles or is it entirely delegated to terraform?

Any reason

vault.Policy

requires the

policy

property as a string instead of an object?

I have a resource (Kubernetes Container Cluster) that won’t allow two properties to update at the same time (node count and version) as the api rejects them as mutually exclusive changes. This is a problem in Terraform as well that I haven’t quite figured a good solution for. Is there a good way in Pulumi to model updating a resource, then coming back in and updating that resource again as a separate task and not compound the changes to that resource with both updates?

Is there a way to see what the diff is when you get

Plan apply failed: inputs to import do not match the existing resource

?

@gentle-diamond-70147 Hi Cameron, I work with Duck over at Vizient. We are building our own pulumi module which wraps up various in house patterns into an easily consumable module. It's written in typescript against Azure. I am currently actively developing this typescript module 'vznpulumi' and trying to spin it up in a known good stack. The vznpulumi module sits at the same directory level as my working stack "hondofoozle". I have imported this local version of the vznpulumi module and the other modules we need like this

import * as aad from '@pulumi/azuread'; import * as pulumi from "@pulumi/pulumi"; import * as azure from "@pulumi/azure"; import * as vzn from '../vznpulumi/src';

But when running pulumi up, I get this wierd error.

Previewing update (vizientinc/sandbox): Type Name Plan Info + pulumi😛ulumi:Stack hondofoozle-sandbox create 1 error + └─ azure:core:ResourceGroup rg create Diagnostics: pulumi😛ulumi:Stack (hondofoozle-sandbox): error: Running program 'C:\work\repos\hondofoozle' failed with an unhandled exception: Error: EBUSY: resource busy or locked, open '\\.\pipe\pulumi1444445712\invoke_req' at Object.openSync (fs.js:448:3) at Object.tryGetSyncInvokes (C:\work\repos\hondofoozle\node_modules\@pulumi\pulumi\runtime\settings.js:163:29) at invokeSync (C:\work\repos\hondofoozle\node_modules\@pulumi\pulumi\runtime\invoke.js:95:36) at Object.invoke (C:\work\repos\hondofoozle\node_modules\@pulumi\pulumi\runtime\invoke.js:84:24) at Object.getSubnet (C:\work\repos\hondofoozle\node_modules\@pulumi\network\getSubnet.ts:37:62) at Object.<anonymous> (C:\work\repos\hondofoozle\index.ts:26:30) at Module._compile (internal/modules/cjs/loader.js:936:30) at Module.m._compile (C:\work\repos\hondofoozle\node_modules\ts-node\src\index.ts:439:23) at Module._extensions..js (internal/modules/cjs/loader.js:947:10) at Object.require.extensions.<computed> [as .ts] (C:\work\repos\hondofoozle\node_modules\ts-node\src\index.ts:44

THere are no other instances of node or pulumi running. It's a fresh reboot. All yarn files match. And I can't find anything about this on the net. Chris Duck is running the same setup and is not having this issue. Have you see this issue?

I’m using a Javascript function for Lambda on Edge, with

runtime: aws.lambda.NodeJS8d10Runtime,

but then received an email from AWS that that runtime will be obsoleted:

your AWS Account currently has one or more Lambda functions using Node.js 8.10, which will reach its EOL at the end of 2019.

So I wanted to switch to then next version but

error TS2551: Property 'NodeJS10dXRuntime' does not exist on type 'typeof import("/.../node_modules/@pulumi/aws/lambda/i...'. Did you mean 'NodeJS6d10Runtime'

Note that

NodeJS10dXRuntime

is still documented here: https://www.pulumi.com/docs/reference/pkg/nodejs/pulumi/aws/lambda/#NodeJS10dXRuntime So I changed to using the string

"nodejs10.x"

which didn’t work

error: aws:lambda/function:Function resource 'language-redirect' has a problem: expected runtime to be one of [dotnetcore1.0 dotnetcore2.0 dotnetcore2.1 go1.x java8 nodejs4.3 nodejs4.3-edge nodejs6.10 nodejs8.10 provided python2.7 python3.6 python3.7 ruby2.5], got nodejs10.x

and then went back to

runtime: "nodejs8.10"

which is the one that will be obsoleted soon. So… what should I do here?

There is some serious jank with Release Channels

for instance there is no way to define the latest version of a Release Channel for the initial create of your cluster with a release channel

you would expect there to be a filter clause on getEngineVersions something like channel, but no...

Ok I found an interesting problem I think

a pulumi refresh doesn't pull stack reference info, so if you share your provider via a stack and your nodes change because of a replacement in a microstack, when you attempt to do a pulumi update in another microstack that depends on the provider from a stack reference you get a timeout error

It would be interesting to let you run a command to update certain stack references

I guess you can do this via --target

so yes that works great, but its a pain in the butt to do it that way...

Hello, I'm getting `System.InvalidOperationException: Expected System.String but got System.Collections.Immutable.ImmutableArray`1[[System.Object, System.Private.CoreLib, Version=4.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e]] deserializing Pulumi.Azure.AppService.AppService.appSettings` after updating to Pulumi 1.6.0

How to delete and start login account all over again

This feels like a dumb question... but am I supposed to commit my stack config files, like

Pulumi.dev.yaml

? Seems like that would all be stored in Pulumi Cloud

This builds on Joe's question: In the stack config files, the encryption salt and the secret are both saved in the file like so:

encryptionsalt: v1:xyz==
config:
  my-project:secretConfig:
    secure: v1:DtxM

Doesn't this pose a security challenge? In what other ways can this be done?

Good Morning. We are working on the CI pipeline for adding a pulumi preview check prior to merging a PR. It currently looks like when selecting --stack on the command line one is not able to do stack init automatically/ if the stack file is already committed to git? I was wondering if anyone had suggestions that work with a s3 backend

Can we please, please, please get an event notification or overall Promise or defer mechanism that fires when a 'pulumi up' is finished? We're jumping through hoops to have integration tests when a pulumi run is done. In most cases we wrap individual tests in 'apply' or 'all' but that doesn't always work. We just had a case where a test would start before a model was assigned to a variable so we couldn't use either approach and had to switch to an event emitter to hold the test back. This seems ridiculous. Why can't we get an overall 'pulumi up/preview/destroy' is done, run your follow-on code now?

Within corporate environment under ADMIN mode Powershell, I was able to login and pulumi whoami recognized username, but pulling down from existing project created previously outside of corporate environment with: > pulumi new aws-typescript -s USERNAME/aws-typescript/dev > > ENTER to log in using your browser > > but opens: "Pulumi doesnt support this version of Microsoft Internet Explorer" please consider upgrading. > > and I dont think I can upgrade at Microsoft IE at clients site??? Please advise if anyone I am sure has faced this situation as well as it is asking for Windows Security iexplore credentials

Can Pulumi and Terraform both exist within a project? So that if someone still wants to work with Terraform and deploy to AWS, and I use Pulumi to deploy to the same AWS cluster? is that ok?

Can I have Pulumi read the state of current infrastructure and use that as a starting point for the first

pulumi up

?