👋 I am cruising around Europe in a campervan 🚌. Garden's developers are here to support you while I'm away but please note they are also busy improving your favorite developer tool so response times may be slower than usual!

When running a integration tests (or workflows running them) I can't find a way to get the logs for the services that the tests are hitting. For instance if I just run an HTTP request in my test all I can see is the error status code, and I need to set up the

garden dev

environment and stream the logs to actually see the stack trace on the server side. It would be really awesome if there was a way to specify which services to capture logs for during a test run. I tried doing something like this:

kind: Workflow
name: integration-tests
steps:
  - command: [deploy, --logs, api]
  - command: [test, integration-tests]

But it gives me an error that specifically it can't stream logs using workflows. Is there a solution I'm missing, or is this something that would be a feature request?

The docs on the website list a config which isn't actually compatible (using hub.docker.com results in 404s) and I'm unable to actually get auth to work with standard dockerhub secrets (index.docker.io)

Hi folks! Since recently some of our developers sporadically hit a generic error

Invalid count value

when deploying with local mode.

$ garden deploy --local bex-os

...
Scanning repository at /Users/xxxx/IdeaProjects/bex-os


Invalid count value
Scanning repository at /Users/xxxx/IdeaProjects/bex-umbrella
Scanning repository at /Users/xxxx/IdeaProjects/bex-umbrella/.garden/sources/project/bexapp--accbe12695
Scanning repository at /Users/xxxx/IdeaProjects/bex-umbrella/.garden/sources/project/bom-app--fc03d112c9
Scanning repository at /Users/xxxx/IdeaProjects/bex-umbrella/.garden/sources/project/bom--493b1ac2ed
Scanning repository at /Users/xxxx/IdeaProjects/bex-os


$ echo $?
1

When deploying without the

--local

flag, the problem does not occur. Most of the time, the problem does not occur. However, when it occurs, it keeps occurring until the next day. I'm still trying to reproduce this issue but so far had no luck, since it never occurs on my machine. All our devs are using OSX - I'm on linux. I grepped the garden codebase for the error string and as well only parts of it but couldn't find any sensible matching occurrence. It might come from a `RangeError`: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Errors/Resulting_string_too_large Has anyone seen this behavior before? Any ideas what the problem could be?

Is

garden-system

namespace deprecated and planned to be removed? As I see garden can work without this ns. I want to deploy some custom resources to this namespace and want to know beforehand what will Garden do with it in future.

Hi I am trying to deploy a helm deployment on our Kubernetes cluster from a Jenkins pipeline to run some integration tests. For this, I use an existing chart which I have tested well while running garden from my machine. But when running in Jenkins, I had to create a new cluster admin role with specific permissions (see snippet attached) that I don't have to use while running locally due to my user having complete cluster access. I created this to provide kubectl access to Jenkins. I am getting an error that has my head scratching

ℹ deploy.gdp-garden [verbose] → [kubernetes-plugin] Status of Deployment dp-ci-1691586975313-gdp is "deploying"

⚠ deploy.gdp-garden → Deployment/dp-ci-1691586975313-gdp: Error: failed to generate container "5619a71419b472824c14778e2632e2ed39f0571748f98e54023cf05c87e231cf" spec: failed to generate spec: no command specified

ℹ deploy.gdp-garden [verbose] → [kubernetes-plugin] Deployment/dp-ci-1691586975313-gdp: Error: failed to generate container "5619a71419b472824c14778e2632e2ed39f0571748f98e54023cf05c87e231cf" spec: failed to generate spec: no command specified

A CMD is present inside the Dockerfile to build the container used in this deployment, so our helm chart does not have any command specified for the deployment, and it has always worked in the past (and when I deployed the chart locally using garden). I could add a command to the manifest or helm values to bypass this error. But I am wondering why there is a difference in behaviors. Am I missing some permission in the ClusterRole?

Hi, our pipeline with garden sometimes take over an hour, is there a way to speed things up for the production pipeline?

We have garden setup for local development purposes and we use cluster-buildkit mode for building docker images. But we want to configure kaniko build mode with the same Garden project for running Garden in CI/CD. Currently, I do not see any possibility to define both build modes in one project and choose one for either of cases. Is it even possible? If yes, how can I do it?

No idea if this is a known issue. I can create a bug ticket if it is not. If you have a Run action (a) that depends on a Run action (b) that depends on a Deploy action, and you’ve run all of these actions before, their statuses get cached so Garden doesn’t need to re-run them — all good so far. Then, if you delete and redeploy the Deploy action, I would want that to invalidate the previous run of both actions (b) and (a) but it does not. You can even force Run action (a) to run and it will error out (internally to the contents of the action, not a Garden error) because Garden will skip over (b) as having been run when in reality it has not been run against the new Deploy.

Hello! We're a small team using Garden to manage and deploy a set of AI workloads and services on clusters in different environments, and it's working really well for us. We have until now used a garden build+test+deploy workflow for all our environments (dev, test and prod) using container Build, container Test and kubernetes Deploy actions. However, we want to implement a GitOps approach for our prod environment, in which Garden is only responsible for building images and rendering k8s manifests, after which the manifests should be handed over to a GitOps tool (we're trying out the Pulumi Kubernetes Operator at the moment). But we're not quite sure what is the most natural way to use Garden to just render and output the k8s manifests without deploying them. I can think of a few hacky ways, but since I imagine this might be a reasonably common use case, I'm wondering if I'm missing something and if there is one or more "natural" patterns for such a workflow using Garden.

I'm kind of using Garden as a replacement for manually scripting builds in either CI/CD scripts (i.e. Gitlab yaml) or in my Dockerfiles. I like being able to run the same thing locally and in CI as it makes the whole cycle much faster; and I like writing the script outside of the Dockerfile since its much easier to debug failing builds when its outside of the Dockerfile and I have to deal less with injecting tokens/keys etc. which is a PITA with docker (and different if you use

docker build

locally, and kaniko remotely). My goal is that I want an action to download/clone another repo, run (in effect)

make build

, and then have the outputs be available to other build tasks in my garden graph. I've run into a couple dead-ends so far though: 1. I tried just using "Build exec", which works if you don't sandbox anything which would be the more desirable way of working. Furthermore, this is not not ideal for running in CI and doesn't work in most cases. First, my CI runner is using the

gardendev/garden-gcloud

image which is good for running garden, but not so much for my own toolchain and may actually be incompatible (for instance this image is on alpine). I could make my own image but then it would be a weird hybrid and hard to maintain. Without that concern it would be better just to run exactly the same build container locally and in CI. 2. So I tried running the builds inside of a Pod in the cluster (whether that is local minikube or remote GKE). The problem here is that there is no Build type for running in an exec-like command in a pod (like

kubernertes-pod

for the Run action type). Using Run kubernetes-pod works, but the semantics around passing data between actions is only really set up for Build action types. I can output an artifact but AFAIK there is no good way to get this into the Build container workspace. So I think my feature request would be Build types for kubernetes-pod and kubernetes-exec. Would this be a difficult thing to add?

Trying to run‘garden .13 on a Mac M1 with minikube - I’m getting ‘exec /rsync-server.sh: exec format error’. Any way to get —sync mode to work on Mac silicon? Tia

Hey guys. Is there an equivalent to generateFiles in the Terraform Deploy schema? My use-case is the following: I have Terraform Deploy actions which pull their source files from another repository (specified in source.repository.url). I would like to use a remote Terraform backend for these actions. Garden sadly does not expose a way to configure Terraform state backends. My workaround with Garden modules was using generateFiles to copy a backend.tf file with some templating, into the computed module sources as follows:

generateFiles:
  - sourcePath: ./backend-file-name
    targetPath: ./path/to/module/root
    resolveTemplates: True

Is there a way to achieve the same outcome with Garden Actions? Thanks in advance!

I was trying the quickstart(https://docs.garden.io/basics/quickstart) on bonsai/windows with docker-desktop and running into this issue running deploy. Any ideas here?

Hi All, I'm looking into introducing garden (I really love the idea) and I'm trying to clear in my head how it fits with the way our repos are structured and our code is deployed so I can try get a POC running. We have multiple backend services spread across a few mono-repos (each repo represents a feature of the product, and has a directory per service that makes up that feature), another repo represents our frontend/UI, and a final repo contains the Jenkins jobs for testing post-merge. Each service has a Dockerfile and builds and pushes a helm chart. Our CD repository contains environment configurations. At a surface level this repo is made up of a directory per env. This env directory contains one large values.yaml, and subdirectories per service to be deployed with a yaml file describing the helm chart to deploy. Questions 1. Given the multi-repo nature of the code, are remote sources the right approach or is there something different I have missed? 2. Given all the code repos are of equal importance, where should the root garden project file have its own repo? It seems either: the CD repo, or a similar decoupled repo should store it, OR each dev team has their own garden project in their repo that lists their dependencies as the other repos?

Hello, I'm trying to get garden working using the

kubernetes

provider. Garden version 0.13.12. I have the

local-kubernetes

working just fine but am having trouble getting it set up for remote. I am using this document as my general guide https://docs.garden.io/kubernetes-plugins/remote-k8s/configure-provider. I setup a cluster in Azure AKS and can connect to it using

kubectl

(e.g.

kubectl get nodes

). I deployed contour, cert-manager, letsencrypt and certificate using helm/kubectl such that I have a valid certificate. My DNS points to the cluster (if I install a service using helm I can connect to it at my URL). I have a container registry created in azure and created a imagePullSecret and put it in the cluster using

kubectl create secret docker-registry

My

project.garden.yaml

looks like: Edit - config removed for length. The issue is when I run

garden deploy --env remote

I get the following message:

ℹ garden               → Running in Garden environment remote.default
ℹ providers            → Getting status...
✖ providers            → Failed processing resolve provider kubernetes (took 1.4 sec). Here is the output:

──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Error: Client network socket disconnected before secure TLS connection was established
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────


Failed resolving one or more providers:
- kubernetes

See .garden/error.log for detailed error message

There does not appear to be any additional information in the logs. I've tried running with the

--logs

flag which appears to run it in an interactive mode, but I am not sure what to do from there. I am not sure where to go from here. Any help would be appreciated. Thanks, Matt

I'm getting an error when trying to pass some local environment variables to buildargs This works, passing in the actual username and key:

buildArgs: {
    GITHUB_API_USER: username,
    GITHUB_API_KEY: api_key
  }

None of the combinations I'm trying work, eg:

buildArgs:
    - GITHUB_API_USER: ${local.env.GITHUB_API_USER}
    - GITHUB_API_KEY: ${local.env.GITHUB_API_KEY}

this results in the error:

Error validating spec for Build type=container name=policies: key .buildArgs must be of type object. Available keys: localId, publishId, targetStage, buildArgs, extraFlags, dockerfile, hotReload)

I'm not sure how I create an object that can be passed to buildargs using local env vars?

I'm having a hard time understanding and reproducing this issue, but I thought I would reach out for tips and to see if anyone else had experienced this. Occasionally garden will fail in container Build actions because it fails to include files for the build context. I've triple checked my

.gardenignore

and my

include

section and there is no doubt that they are included. And indeed the same configuration works at other times. Its always the same files interestingly

Makefile

, and e.g.

dev/containers/tests.Dockerfile

. I've tried renaming the files, clearing the cache and build envs, running with

garden --force-refresh build thing --force

, nuking the entire project

.garden

directory, reinstalling garden and nuking

~/.garden

and nothing seems to work except waiting (usually the next day...). The last bit makes me think that there is something wrong with timestamp checking or something. This is pretty breaking and basically halts my work until garden decides to fix itself. Any ideas at all?

Hey, After spending some time messing around with garden.io I have not been able to get php-fpm to work. I tried configuring both an external ingress and trying to configure the default ingress, but I have been unsuccessful in all my attempts to get my laravel application to serve. My current garden.yml for php-fpm is as follows: kind: Build type: container name: php-fpm --- kind: Deploy description: The php container name: php-fpm type: container build: php-fpm spec: ports: - name: https protocol: TCP containerPort: 9000 sync: paths: - target: /var/www/html source: ../../src ingresses: - annotations: nginx.ingress.kubernetes.io/rewrite-target: /index.php?$2 nginx.ingress.kubernetes.io/add-base-url: "true" nginx.ingress.kubernetes.io/backend-protocol: "FASTCGI" nginx.ingress.kubernetes.io/fastcgi-index: "index.php" nginx.ingress.kubernetes.io/fastcgi-params-configmap: "php-fpm-fastcgi-configmap" # Add other annotations as required path: / hostname: dev-vytal.ap.ngrok.io port: https Any help pointing me in the right direction for how to properly get this setup for PHP-FPM would be greatly appreciated. I am relatively new to kubernetes and garden.io in general, so my apologies in advance if I am overlooking something crucial. Thanks in advance, Price

Hi again! Using Garden 0.13.12, I'm trying to isolate an individual terraform deploy action so that I correctly order cleanup operations and their dependencies. When trying to pass the name of the terraform deploy action to the commands: -

garden plugin terraform destroy-action fsx-csi-driver-tf

-

garden plugin terraform apply-action fsx-csi-driver-tf

-

garden plugin terraform plan-action fsx-csi-driver-tf

I receive the following error:

Unexpected positional argument "fsx-csi-driver-tf" (expected only plugin and command)

Based upon the CLI output for the commands, I would expect to pass an argument with the name of the action to the command.

Runs terraform apply for the specified terraform Deploy action, with variables automatically configured as inputs. Use the action name as first argument, followed by any arguments you want to pass to the command. If necessary, terraform init is run first.

Am I using this wrong or could there possibly be an issue with the CLI parser not accepting positional arguments from the plugin? Thanks, Jess

Welcome!

acoustic-dusk-48036

joined #🌱｜help-and-getting-started.

Hi! I'm having a issue when try to uso de command garden plugins terraform apply-action -- platform . The error is the following: Failed extracting zip archive https://releases.hashicorp.com/terraform/0.14.7/terraform_0.14.7_darwin_amd64.zip Invalid checksum from https://releases.hashicorp.com/terraform/0.14.7/terraform_0.14.7_darwin_amd64.zip (got bd4afbb92cfc99f3f7e81412536e1aa9bafd6544a87454286d9e9f6ab446179a) My OS is MacOs, Any idea?

Hey folks, I'm not sure if this is a known problem going from 0.12 to 0.13 with a documented fix somewhere, but I can no longer run

garden deploy <thing>

without it failing with the error

Lock file is already being held

reported for random modules. This is consistent with every run -- I haven't been able to run

garden deploy

at all since upgrading to 0.13. I haven't yet gotten around to converting modules to actions since I wanted to see first if I could ask my team to upgrade to 0.13 while we convert things over time. I think 0.13 should work with modules to enable this, but we have too many modules (47 as of this writing) to test converting things piecemeal if I'm getting lock errors every run. I tried reinstalling Garden as well (cleaning out the existing .garden directories and starting clean), but that hasn't helped. Anyone know if there's some obvious thing I'm missing here? A lock file I should be looking for to clean up? Should it be treated as a bug? Thanks for any suggestions or time spent thinking about this.

Is there an example of creating a service account as part of the deploy? The service-account creation is defined outside of the repo and longer-term I might look to use that definition, but manually running

kubectl create sa testaccount --namespace=test-dev

is enough to get the service to start for now. For some more context my service deployment is defined using helm charts, and it's possible to override the value of serviceAccountName to get the service to expect a different service account.

Welcome!

gifted-thailand-88875

joined #🌱｜help-and-getting-started.

I have two helm template files:

api-deployment.yaml
worker-deployment.yaml

Initially I'm trying to only deploy the worker, so I added an exclude clause in my garden.yaml for the helm deployment:

exclude:
  - 'api-deployment.yaml'

This works at excluding the file; I can see in the debug logs

Found 5 files in Deploy action worker-deploy root /Users/mark/workspace/policies/.k8s/helm

, where previously it found 6. However, the deployment still attempts to deploy the api application. I'm assuming this is something to do with how Helm works and is called under the hood. Is there a way to do what I'm trying to do? Maybe it's just too far outside of a reasonable usecase?

Hello, Is there a way to overwrite the node heap size? I am getting JavaScript heap out of memory issues

Hey y'all! Wanted to check if what I'm aiming to do here is possible or if maybe I should try to go to a different solution (maybe using localPort) https://discord.com/channels/817392104711651328/1120389258683883692/1155927557963776131 What I want to achieve is not having to do a port-forward of my postgres instances when I'm developing locally but be able to reach them through Ingress. I think maybe the localPort would work as well as I could have an specific port for it, but then I would need multiple because I have mutliple postgres instances running. Thanks ❤️

Hey everyone, I've setup a garden project with a deployment registry & k3d. The project buildMod is setup on

cluster-buildkit

and deployment are working. But build does not work anymore since i've switch from my ArchLinux to Mac M1 I suppose the platform build target is the problem. If you have something in mind that could help me.

is it possible to add metadata annotations to a namespace through garden? or do we have to create the namespace first and then do a garden deploy