version is missing from previous step

[robertskworld]

Hello

I have the latest version of concourse deployed on a Kubernetes cluster without persistence.
(K8s Image: concourse/concourse:7.11.2)
I want to deploy the pipeline via your ansible resource type, through this yaml, with fly:

fly set-pipeline /target test /pipeline ansible-playbook2 /config ./pipeline-ansible2.yaml

---

resource_types:

• name: ansible
  type: docker-image
  source:
  repository: troykinsella/concourse-ansible-playbook-resource

resources:

• name: ansible
  type: ansible
  source:
  verbose: v

jobs:

• name: ansible-playbook
  plan:
  • get: ansible

---

fly deployment + resource type check - will be always successful.
But when I check the resource - I always get this error:

fly check-resource /target test /resource:ansible-playbook2/ansible
checking ansible-playbook2/ansible in build 4381
initializing check: ansible
initializing check: ansible
selected worker: concourse-worker-1
version is missing from previous step
version is missing from previous step
errored

I tried setting all possible combinations for version & tag key in source type and source too - no luck.
In your ansible playbook container, I can see that the container is really trying to output the version information:

ansible-test:/opt/resource# ls -all
total 24
drwxr-xr-x 3 root root 4096 Jan 17 18:42 .
drwxr-xr-x 1 root root 4096 Jan 17 18:42 ..
-rwxr-xr-x 1 root root 20 Jul 3 2023 check
-rwxr-xr-x 1 root root 34 Jul 3 2023 in
drwxrwxr-x 3 root root 4096 Nov 20 19:49 lib
-rwxr-xr-x 1 root root 276 Jul 3 2023 out
ansible-test:/opt/resource# cat in
#!/bin/sh
printf '{"version":{}}'
ansible-test:/opt/resource#

When I try another ansible resource type clone (e.g. - argon/concourse-ansible-playbook-resource) - everything works immediately.
This link also solves something similar - concourse/registry-image-resource#316 - but it didn't help me either.

Pls, where and how should I set it so that the resource type check will be successful?

[troykinsella]

Hi there! This resource is put-only, so a get, regardless of this error, would be a no-op.
If you have a playbook that you're trying to run from a pipeline, you can pull in a git resource, for example, and then invoke it via a put:

- name: run-my-playbook
  steps:
  - get: master # your git repo
  - put: ansible
    params:
      check: true
      diff: true
      inventory: inventory/some-hosts.yml
      playbook: site.yml
      path: master

Hope that helps. Let me know if you had something else in mind.

[robertskworld]

Thank you very much for your answer.
I would like to draw attention to the fact that my problem is not in the implementation of put/get via the job - but in the initialization check of the resource type itself, which is done before the implementation/build, as seen in the printscreen.

So for clarity:

• I also tried it with the put option - the same result
• here is my procedure, with the described result:

-> cat pipeline-ansible.yaml

-> fly destroy-pipeline /target ace-test /p deploy-ansible-playbook
!!! this will remove all data for pipeline deploy-ansible-playbook

are you sure? [yN]: y
deploy-ansible-playbook does not exist

-> fly set-pipeline /target ace-test /pipeline deploy-ansible-playbook /config ./pipeline-ansible.yaml
resources:
resource ansible has been added:

• icon: ansible

• name: ansible

• source:

• debug: true

• verbose: vv

• type: ansible

  resource git has been added:

• icon: github

• name: git

• source:

• branch: master

• password: ((git-acecloud-interny.gitpswd))

• skip_ssl_verification: true

• uri: https://gitlab/........./linux-playbooks

• username: ((git-acecloud-interny.gituser))

• type: git

resource types:
resource type ansible has been added:

• name: ansible
• source:
• repository: troykinsella/concourse-ansible-playbook-resource
• tag: latest
• type: docker-image

jobs:
job deploy-ansible-playbook has been added:

• name: deploy-ansible-playbook
• plan:
• • get: git
• • params:

• inventory: _play.lin.node_exporter/inventory/hosts
  

• path: git
  

• playbook: _play.lin.node_exporter/_ansible_playbook.linux.prometheus.yml
  

• put: ansible

pipeline name: deploy-ansible-playbook

apply configuration? [yN]: y
pipeline created!
you can view your pipeline here: https://concourse............../pipelines/deploy-ansible-playbook

the pipeline is currently paused. to unpause, either:

• run the unpause-pipeline command:
  C:\sklad\concourse_cli\fly.exe -t ace-test unpause-pipeline -p deploy-ansible-playbook
• click play next to the pipeline in the web ui

[troykinsella]

Do you have a private Docker registry in place, perhaps? If so, are you able to delete the latest tag and re-pull? Can you fetch the image directly from a Concourse worker?

What happens when you fly check-resource-type --resource-type deploy-ansible-playbook/ansible?

How about a fly clear-versions?

Sorry if these suggestions are obvious. I've not encountered this error before.

[robertskworld]

Hello,

I had to change the pipeline a bit, but the result is the same.
I don't use any private repo, I call DockerHub directly.
As I wrote, the error is reported by the resource and not the resource type check.
I increased the debug level, but from the monitoring results - I don't see anything new.
Just checking the source gives me an error, but I can't tell if it's a reason.
I could not find out how to fetch the image from the worker node, if you advise, I will do it. But I have a feeling that it will only happen when the resource's init check passes.

FYI - fly CLI is on windows
fly --version
7.11.2

[robertskworld]

Hello,

FYI:
when I use the version 2.2.2 of your source type - everything works, is OK.
So the problem will be around the changes of the last version.

One more question please:
Is it possible to change the version of ansible, or is it always static according to the version of resource type?

[troykinsella]

I've rebuilt the image and pushed a fresh latest tag. Can you please check if the issue is resolved?

You could probably install a specific version of Ansible upon every job run, as per this explanation for additional modules: https://github.com/troykinsella/concourse-ansible-playbook-resource?tab=readme-ov-file#custom-setup-commands
But, the overhead of doing so might be undesirable, and if so, I think you're looking at extending this image and installing the things that you require.

[robertskworld]

I confirm that this change removed the notified error message. Dňa pi 1. 3. 2024, 19:51 Troy Kinsella ***@***.***> napísal(a):

…

I've rebuilt the image and pushed a fresh latest tag. Can you please check if the issue is resolved? You could probably install a specific version of Ansible upon every job run, as per this explanation for additional modules: https://github.com/troykinsella/concourse-ansible-playbook-resource?tab=readme-ov-file#custom-setup-commands But, the overhead of doing so might be undesirable, and if so, I think you're looking at extending this image and installing the things that you require. — Reply to this email directly, view it on GitHub <#31 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BGGV2SJE33V3L7SYH2RJAJTYWDEZVAVCNFSM6AAAAABDMJCOBSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZTG42DKNBZGE> . You are receiving this because you authored the thread.Message ID: <troykinsella/concourse-ansible-playbook-resource/issues/31/1973745491@ github.com>