How I can trigger a job from a previous one?

[akaibukai]

Hello,
So I have 3 jobs, A, B and C.
Basically I want that when A pass, it triggers B and C that will run in parallel, but after A.
I saw there's a passed option but it's applicable when you have a ressource only (or I miss something?)
Is it possible to do that?
Here is a sample pipeline:

---
jobs:
- name: job-a
  plan:
  - task: simple-task
    config:
      platform: linux
      image_resource:
        type: registry-image
        source: { repository: busybox }
      run:
        path: sh
        args:
          - -exc
          - ls -la; echo "Created a file on $(date)" > ./files/created_file
      outputs:
        - name: files

- name: job-b
  plan:
  - task: consume-the-file
    config:
      platform: linux
      image_resource:
        type: registry-image
        source: { repository: busybox }
      inputs:
        - name: files
      run:
        path: cat
        args:
          - ./files/created_file

- name: job-c
  plan:
  - task: consume-the-file
    config:
      platform: linux
      image_resource:
        type: registry-image
        source: { repository: busybox }
      inputs:
        - name: files
      run:
        path: cat
        args:
          - ./files/created_file

Thanks!

[suhlig]

Things are happening in Concourse because there is a new version of a resource (or when jobs are started manually). So if you want B to run after A, then A needs to put a new version of some resource (which you'll refer to in the passed section).

If there is no natural new version of something A produces, then you could invent something. Conceptually, I think of that as publishing an event. It could, for instance, be a new or changed file in a git repo or in an S3 bucket.

But don't go there too early - usually this is a chance to discover a hidden concept that, when uncovered, improves your build architecture a lot.

[akaibukai]

Hi thanks for the message..

Conceptually, I think of that as publishing an event.

I agree with that and this is how I also understand.

It could, for instance, be a new or changed file in a git repo or in an S3 bucket.

But I found it too cumbersome to trigger a job by tinkering with an external storage or a repo

then A needs to put a new version of some resource (which you'll refer to in the passed section)

Is it possible to do that with a local file system change?
Job A touch a file for instance..
Just using a regular file output on A and file input on B and C?

Edit: It seems what I was refering as inputs and outputs are actually referred in Concourse as Tasks inputs and Tasks outputs.

It seems there is no such thing for Jobs.

Does it mean I cannot share artifact just by using the file system?

[suhlig]

Correct. The only way to share data between jobs are resources.

[akaibukai]

Thanks for the confirmation..
Do you think it exists a resource like s3 but that uses a local directory (or a docker volume maybe)?

[suhlig]

Such a thing cannot exist, because jobs run on different workers without sharing a common file system.

Resource inputs and outputs are made available to a job as volume mounts. So you will need a resource.