Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Copy output from a previous task, resource is not used ? #1604

Closed
eddycharly opened this issue Nov 22, 2019 · 12 comments
Closed

Copy output from a previous task, resource is not used ? #1604

eddycharly opened this issue Nov 22, 2019 · 12 comments
Labels
area/api Indicates an issue or PR that deals with the API. kind/question Issues or PRs that are questions around the project or a particular feature triage/duplicate Indicates an issue is a duplicate of other open issue.

Comments

@eddycharly
Copy link
Member

eddycharly commented Nov 22, 2019

When a task in a pipeline uses output from a previous task, the resource element is ignored ?
Do the output and input have to share the same name for it to work ?

If i trust the code below :

var stepSourceNames []string
if pt.Resources != nil {
for _, pipelineTaskInput := range pt.Resources.Inputs {
if pipelineTaskInput.Name == name {
for _, constr := range pipelineTaskInput.From {
stepSourceNames = append(stepSourceNames, filepath.Join(storageBasePath, constr, name))
}
}
}
}

If the output is named foo in the previous task and the input is named bar in the current task, it looks like it will try to copy from a path like bucket/previous-task/bar instead of bucket/previous-task/foo.

@vincent-pli
Copy link
Member

@eddycharly
Yes, based on current implementation, the next task has no idea what's the name of output resource of the previous task, only know the name of previous PipelineTask.

@eddycharly
Copy link
Member Author

@vincent-pli what's the reasoning behind this ? previous task output and current task input must match ?

So there is no way to take an output of a task and use it as input in another task whatever their names ?

The examples in the docs are confusing https://github.com/tektoncd/pipeline/blob/master/docs/pipelines.md#from. Not sure why resource and from are set here.

    inputs:
      - name: image
        resource: my-image
        from:
          - build-app

@vincent-pli
Copy link
Member

I think the reason is the pipelinerun will be divide to several taskrun in runtime, each taskrun has limited knowledge of others.

The resource in the sample means pipelineresource, it's a refer to a real crd named pipelineresource, has no relationship with the location of the resource :)

@eddycharly
Copy link
Member Author

eddycharly commented Nov 23, 2019

Each taskrun has no knowledge at all of other taskruns. The pipeline creates/orchestrates the taskruns, making the glue between them.

I am trying to understand if an output of a task can be copied as the input of another task, regardless of the output and input names.

For the copy to work, it looks like the name of the input and output have to be the same, it makes the tasks difficult to reuse as each task name their inputs/outputs independently.

Could we improve this mechanism to allow copying ouput a from one task to input b of another task ?

@vdemeester
Copy link
Member

I think this is related/duplicate of #1002. This is (imho) one of the limitation of the PipelineResource (as currently designed) and one of the reason why PipelineResource are (so far) not considered to move to beta until they are redesigned (see #1369 and design doc)

@eddycharly
Copy link
Member Author

eddycharly commented Nov 25, 2019

thanks @vdemeester
i don't have access to the google drive.
i was thinking about it, why not introduce an implicit resource, something like taskstorage to be used by the pipeline ?
from the task point of view it would be a storage, it wouldn't change anything.
from the pipeline point of view, it could be used to pass the taskstorage to the tasks inputs/outputs.

the taskstorage would take care of copying from/to the storage as configured (pvc, gcs or s3).
it wouldn't be too difficult to implement i guess as all the code is already present.

@vdemeester
Copy link
Member

@eddycharly the current "idea" is to separate this concern from the PipelineResource, see #1438

@vdemeester
Copy link
Member

You need to be part of tekton-dev@ or tekton-users@ to be able to have access to those design document, see here

@vdemeester
Copy link
Member

/kind question
/area api
/triage duplicate

@tekton-robot tekton-robot added kind/question Issues or PRs that are questions around the project or a particular feature area/api Indicates an issue or PR that deals with the API. triage/duplicate Indicates an issue is a duplicate of other open issue. labels Dec 9, 2019
@bobcatfish bobcatfish added this to Needs triage in Tekton Pipelines Feb 6, 2020
@dibyom
Copy link
Member

dibyom commented Mar 18, 2020

Workspaces are in and can solve this problem: https://github.com/tektoncd/pipeline/blob/master/docs/workspaces.md

@vdemeester ok to close this?

@dibyom dibyom moved this from Needs triage to Waiting for Response (from author) in Tekton Pipelines Mar 18, 2020
@eddycharly
Copy link
Member Author

@dibyom two task running on two different can share a workspace ?

@eddycharly
Copy link
Member Author

Closing as it is probably not relevant anymore in regards of workspaces.

Tekton Pipelines automation moved this from Waiting for Response (from author) to Closed May 14, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
area/api Indicates an issue or PR that deals with the API. kind/question Issues or PRs that are questions around the project or a particular feature triage/duplicate Indicates an issue is a duplicate of other open issue.
Projects
No open projects
Tekton Pipelines
  
Closed
Development

No branches or pull requests

5 participants