Error: Failed to init clients #231

mikzuit · 2023-02-20T17:25:20Z

Hi @oboukili could this be related to sh scripts, LF CRLF or argocd provider authentication?

Setup:
Terraform : 1.0.6
Argocd terraform provider: 4.3.0

Config files:

I execute this script in order to add my agent ssh then use it in private repos.
ssh-init.sh

#!/bin/sh
set -e

eval `ssh-agent -s` > /dev/null 2>&1

ssh-add -k  "~/.ssh/id_rsa" > /dev/null 2>&1

main.tf

resource "null_resource" "init" {
  provisioner "local-exec" {
    command = trimspace(file("${path.module}/scripts/ssh-init.sh"))
  }
}


provider "argocd" {
  username    = "admin"
  password    = "password" 
  context     = "my-context"
  server_addr = "ip_address"
  insecure    = true
  kubernetes {
    config_path            = "~/.kube/config"
    cluster_ca_certificate = "from cluster config" 
    exec {
      api_version = "client.authentication.k8s.io/__internal" # also tried  with v1 or v1beta
      args        = ["login", "ip_address", "--grpc-web", "--username", "admin", "--password", "password", "--insecure"]
      command     = "argocd"
    }
  }
}


resource "argocd_project" "deployed" {
}

Regular output:

│ Error: Failed to init clients
│
│   with argocd_project.deployed,
│   on main.tf line 56, in resource "argocd_project" "deployed":
│   56: resource "argocd_project" "deployed" {
│
│ EOF

... 

same error for each argocd resource

TF_LOG=Debug output:

│ Error: Failed to init clients
│
│   with argocd_project.deployed,
│   on main.tf line 56, in resource "argocd_project" "deployed":
│   56: resource "argocd_project" "deployed" {
│
│ EOF

... 

same error for each argocd resource

2023-02-20T16:55:10.110Z [DEBUG] provider.stdio: received EOF, stopping recv loop: err="rpc error: code = Unavailable desc = transport is closing"
2023-02-20T16:55:10.115Z [DEBUG] provider: plugin process exited: path=.terraform/providers/registry.terraform.io/oboukili/argocd/4.3.0/linux_amd64/terraform-provider-argocd_v4.3.0 pid=24320
2023-02-20T16:55:10.115Z [DEBUG] provider: plugin exited

Questions:

how do I check/test my argocd provider connection ?
how can improve my debug argocd provider connection?
Assuming I get error on first resources created, should I consider that argocd provider connection is enabled?
could this be related to using a script with bash/sh ?
Could this be related in any way to?:
Failed to init client: kubernetes configuration invalid #134

I have used the env variables KUBE_CONFIG_PATH , KUBECONFIG as suggested there.

Other things I tried:

Login directly on dashboard same credentials for admin are working
Login from terminal using argocd-cli same credentials for admin are also working

UPDATE: (solution)
This is the minimun login with cli (--plaintext) to avoid TLS error warning message):

argocd login IP_ADDRESS --username admin  --password $argoPass --plaintext

Agocd providers does have that option as plain_text which SOLVES the problem.
Thanks for the help @onematchfox @Diliz

Thanks

The text was updated successfully, but these errors were encountered:

onematchfox · 2023-02-23T06:59:45Z

I haven't used the provider alongside a kubernetes configuration block, so I don't know how much value I can add to this conversation. However, I suspect this is related to how you have exposed your ArgoCD instance, if at all.

Login directly on dashboard same credentials for admin are working

Login from terminal using argocd-cli same credentials for admin are also working

How have you gone about doing these checks? In particular, which parameters are you passing to the argocd login command? Do these match the provider configuration? At this point, I guess that perhaps you should be setting port_forward and port_forward_with_namespace on the provider configuration.

Diliz · 2023-02-23T14:21:07Z

Hello there,

Was having the same issue on an openshift cluster with their argo installation (which is named openshift-gitops), got another error when adding the port_forward options:

Error: Failed to init clients
│ 
│   with argocd_project.argo_project,
│   on providers.tf line 106, in resource "argocd_project" "argo_project":
│  106: resource "argocd_project" "argo_project" {
│ cannot find pod with selector:
│ [app.kubernetes.io/name=argocd-server]

I think this is related to: argoproj/argo-cd#10200 (which I created)
Is there any way to target another argocd-server name (by giving the provider a server name or this kind of stuff for custom argo implementations) ?

If it's related to the issue I mentionned, a pull request is still waiting for the merge here: argoproj/argo-cd#11297

EDIT:
yeah in my case (openshift-gitops argo installation), the issue is in the argocd apiclient as well
@mikzuit if the port-forward options are not working, are you using a default argocd installation? (like official helm chart or else)

mikzuit · 2023-02-23T17:35:20Z

I haven't used the provider alongside a kubernetes configuration block, so I don't know how much value I can add to this conversation. However, I suspect this is related to how you have exposed your ArgoCD instance, if at all.

Login directly on dashboard same credentials for admin are working

Login from terminal using argocd-cli same credentials for admin are also working

How have you gone about doing these checks? In particular, which parameters are you passing to the argocd login command? Do these match the provider configuration? At this point, I guess that perhaps you should be setting port_forward and port_forward_with_namespace on the provider configuration.

Sorry @onematchfox I will answer from memory, I will come back when have more precise data.

I am using argocd loadbalancer option(not port forward or ingress) .
I am able to connect with admin and password on Web Ui. When login using argocd cli I have used :

argocd login IP_ADDRESS --username admin  --password $argoPass

Also

argocd login IP_ADDRESS --insecure --grpc-web--username admin  --password $argoPass

I will say also tried with a ip address:port

Thanks

mikzuit · 2023-02-23T17:39:00Z

Hello there,

Was having the same issue on an openshift cluster with their argo installation (which is named openshift-gitops), got another error when adding the port_forward options:
Error: Failed to init clients
│ 
│   with argocd_project.argo_project,
│   on providers.tf line 106, in resource "argocd_project" "argo_project":
│  106: resource "argocd_project" "argo_project" {
│ cannot find pod with selector:
│ [app.kubernetes.io/name=argocd-server]
I think this is related to: argoproj/argo-cd#10200 (which I created)
Is there any way to target another argocd-server name (by giving the provider a server name or this kind of stuff for custom argo implementations) ?

If it's related to the issue I mentionned, a pull request is still waiting for the merge here: argoproj/argo-cd#11297

EDIT:
yeah in my case (openshift-gitops argo installation), the issue is in the argocd apiclient as well
@mikzuit if the port-forward options are not working, are you using a default argocd installation? (like official helm chart or else)

Hi @Diliz I am indeed using a argocd default installation from this repo , I will try port forward option ASAP

onematchfox · 2023-02-24T06:40:32Z

Hi @Diliz I am indeed using a argocd default installation from this repo , I will try port forward option ASAP

@mikzuit, the provider should be configured using the same parameters as what works when you use argocd login. If you have exposed ArgoCD via a load balancer then there should be no need to specify/set the context and kubernetes attributes on the provider (or the port forwarding options for that matter too). Just use the same configuration as what works when running the CLI. E.g.:

provider "argocd" {
  server_addr = "IP_ADDRESS"
  username    = "admin"
  password    = "password"
  insecure    = true
  grpc_web    = true
}

onematchfox · 2023-02-24T06:50:40Z

Is there any way to target another argocd-server name (by giving the provider a server name or this kind of stuff for custom argo implementations) ?

@Diliz, I'm afraid the provider uses the argo-cd module just as the argocd CLI does. Right now, there is no option to specify the server name on the API client, so there's nothing we can do here, either. This can be looked at once the PR you mentioned has been merged and released.

Edit: It might also be worth asking if there is a reason you cannot expose the ArgoCD server. If you do so, then, as per my comment above, there is no need to use port-forwarding to connect to it. You can simply configure the provider to connect to whatever DNS/port the server is exposed on.

mikzuit closed this as completed Feb 26, 2023

MarshallOfSound mentioned this issue Jul 5, 2024

feat: add support for customer argocd api server name #405

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Error: Failed to init clients #231

Error: Failed to init clients #231

mikzuit commented Feb 20, 2023 •

edited

Loading

onematchfox commented Feb 23, 2023

Diliz commented Feb 23, 2023 •

edited

Loading

mikzuit commented Feb 23, 2023 •

edited

Loading

mikzuit commented Feb 23, 2023 •

edited

Loading

onematchfox commented Feb 24, 2023

onematchfox commented Feb 24, 2023 •

edited

Loading

Error: Failed to init clients #231

Error: Failed to init clients #231

Comments

mikzuit commented Feb 20, 2023 • edited Loading

onematchfox commented Feb 23, 2023

Diliz commented Feb 23, 2023 • edited Loading

mikzuit commented Feb 23, 2023 • edited Loading

mikzuit commented Feb 23, 2023 • edited Loading

onematchfox commented Feb 24, 2023

onematchfox commented Feb 24, 2023 • edited Loading

mikzuit commented Feb 20, 2023 •

edited

Loading

Diliz commented Feb 23, 2023 •

edited

Loading

mikzuit commented Feb 23, 2023 •

edited

Loading

mikzuit commented Feb 23, 2023 •

edited

Loading

onematchfox commented Feb 24, 2023 •

edited

Loading