[JENKINS-37790] abort on pipeline with parallel not succeeding / providing feedback on nodes

Jenkinsfile is creating ~24 sub-jobs and running 5 of them at a time in parallel.

Each sub-job is running up to 5 VMs that need to be cleaned up on abort to make room for the next jobs.

The only time I have seen abort work on the first click is when all of the jobs are in the queue and none of them are running. Every other time, the console output has given me "Click here to forcibly terminate running steps" and later "Click here to forcibly kill entire build", and if i don't click those links, the job will stick around for days.

In this instance, i've created two support bundles, at the times indicated.

When I went to manually clean up the VMs, i found that 2 out of 5 of the workspaces had been successfully cleaned up before I got there, but all 5 workspaces look the same from the POV of the console output...they all look like they failed to abort.

=============================================================

Aborted by LeahKlearman

...............................................................................................................................................................................................................................................................................................................................................Sending interrupt signal to process

..........................................................Sending interrupt signal to process

Sending interrupt signal to process

Sending interrupt signal to process

Sending interrupt signal to process

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[looks like all the queued sub-jobs have been successfully canceled here]

Click here to forcibly terminate running steps

[clicked "forcibly terminate running steps" now and created bundle support_2016-08-29_20.44.15]

Terminating sh

Terminating sh

Terminating sh

Terminating sh

Terminating sh

[output from one of the jobs]

Click here to forcibly kill entire build

[clicked "forcibly kill entire build" now and created support_2016-08-29_20.45.14]

Hard kill!

=============================================================

---
Originally reported by klrmn, imported from: abort on pipeline with parallel not succeeding / providing feedback on nodes

• status: Open
  
• priority: Major
  
• component(s): workflow-cps-plugin, workflow-durable-task-step-plugin
  
• resolution: Unresolved
  
• votes: 6
  
• watchers: 10
  
• imported: 2025-12-07
  

Raw content of original issue

Jenkinsfile is creating ~24 sub-jobs and running 5 of them at a time in parallel.

Each sub-job is running up to 5 VMs that need to be cleaned up on abort to make room for the next jobs.


The only time I have seen abort work on the first click is when all of the jobs are in the queue and none of them are running. Every other time, the console output has given me "Click here to forcibly terminate running steps" and later "Click here to forcibly kill entire build", and if i don't click those links, the job will stick around for days. 

In this instance, i've created two support bundles, at the times indicated.


When I went to manually clean up the VMs, i found that 2 out of 5 of the workspaces had been successfully cleaned up before I got there, but all 5 workspaces look the same from the POV of the console output...they all look like they failed to abort.


=============================================================

Aborted by LeahKlearman


...............................................................................................................................................................................................................................................................................................................................................Sending interrupt signal to process

..........................................................Sending interrupt signal to process

Sending interrupt signal to process

Sending interrupt signal to process

Sending interrupt signal to process

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[Pipeline] // node

[looks like all the queued sub-jobs have been successfully canceled here]


Click here to forcibly terminate running steps

[clicked "forcibly terminate running steps" now and created bundle support_2016-08-29_20.44.15]

Terminating sh

Terminating sh

Terminating sh

Terminating sh

Terminating sh


[output from one of the jobs]


Click here to forcibly kill entire build

[clicked "forcibly kill entire build" now and created support_2016-08-29_20.45.14]

Hard kill!

=============================================================

• environment: Jenkins 2.7.1, pipeline plugins up to date as of 8/26/16

2 attachments

- [support_2016-08-29_20.44.15-redacted.zip](https://issues.jenkins.io/secure/attachment/33765/support_2016-08-29_20.44.15-redacted.zip)
- [support_2016-08-29_20.45.14-redacted.zip](https://issues.jenkins.io/secure/attachment/33766/support_2016-08-29_20.45.14-redacted.zip)

Start by reviewing the workflow-cps-plugin and workflow-durable-task-step-plugin components, then use the reported Jenkinsfile parallel-job setup and the two support bundles to investigate abort behavior. Done means running parallel branches receive consistent abort feedback and terminate their running steps and workspaces without requiring the forced termination links.

Written by the indexing model from the issue text.