Bug 2135399

Summary:	[Pipelines as Code] Debugging pipeline tasks is cumbersome
Product:	Red Hat OpenShift Pipelines	Reporter:	Xander Soldaat <xsoldaat>
Component:	pipelines	Assignee:	Chmouel Boudjnah <cboudjna>
Status:	NEW ---	QA Contact:	Ruchir Garg <rgarg>
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	unspecified	CC:	cboudjna, kbaig, nikthoma, pgarg, rbehera, sashture
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:		Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Xander Soldaat 2022-10-17 14:02:35 UTC

Description of problem:
Debugging tasks for pipelines is a painful process that takes much more effort than it should. For example, when a pipeline definition is incorrect because a task cannot be resolved, the errors are in the controller, and not in the run output, or GitHub check.
The PaC pipeline disappears from the pull request checks in GitHub, making it seem like everything finished correctly, to an unassuming user.

The logs should be gathered in some fashion, and presented to the user in a uniform manner, and not for them to have to go on a wild goose chase to scour the various places to look for breadcrumbs.

A more user-centric error reporting mechanism will make adoption a much smoother experience.


Version-Release number of selected component (if applicable):


How reproducible:
This is a general observation, and based on customer feedback.

Steps to Reproduce:
1.
2.
3.

Actual results:

Expected results:

Additional info:

Comment 1 Chmouel Boudjnah 2022-10-18 08:32:42 UTC

Hello,

We have a few improvement on the way for the next release, making it easy to know if something bad has hapenned :

- We have fixed a bug to display the "Pipeline definition failures" bad yaml or otherwise, to display in the Github Checks API.
- We are now storing the errors in the user namespace, so user can do a kubectl get events -n ${THEIR_REPO_NAMESPACE} 
  and they would get the parsing errors.
- We are improving our tkn pac CLI to display errors when a failure is detected in the pipelinerun attached to the repo.

I hope this will help,

Comment 2 Chmouel Boudjnah 2022-10-27 09:50:08 UTC

Jira link of the stories. Most of them have been implemented already upstream : 

https://issues.redhat.com/browse/SRVKP-2563