Migrate 'ProcessTreeKiller' from wiki #3321

[bigdevlarry]

Fix #3321

Description

I migrated the ProcessTreeKiller from Wiki to the Destination on jenkins.io: Using Jenkins -> Long-running agent processes where I created an adoc file to copy the converted Ascii format text

[oleg-nenashev]

Thank you for this contribution! It is much appreciated.

Are you doing it as a part of the Jenkins UI/UX Hackfest? If so, please consider reporting this contribution here so that we could facilitate reviews and share the story 👍

[MarkEWaite]

This pull request mixes multiple changes into a single pull request. It needs to be reworked.

@LarrySul I'm happy to host an office hours session or to have a working session in the Gitter chat channel. This is a common challenge for technical writers and it would be good to capture the way to resolve it in a chat session or in an office hours session

[oleg-nenashev]

I suggest cleaning up the obsolete terminology and deprecated features

[bigdevlarry]

@oleg-nenashev @MarkEWaite @markyjackson-taulia been able to resolve all conversations and made an update . Looking forward to your feedback

[MarkEWaite]

See the comments in jenkinsci/workflow-durable-task-step-plugin#39 for the details that I've tried to reduce to this phrasing.

I would love to have comments or corrections on my updated phrasing from @bitwiseman , @dwnusbaum , or @jglick. I believe I've stated it correctly, but they may see mistakes in my additions.

Suggested change

	In case of Jenkins Pipeline use `+JENKINS_NODE_COOKIE+` instead of `+BUILD_ID+`
	In case of Jenkins Pipeline use `+JENKINS_NODE_COOKIE+` instead of `+BUILD_ID+`.
	Parallel stages of the Jenkins Pipeline may run on the same agent and will use the same `+BUILD_ID+` within a specific Pipeline run.
	If `+BUILD_ID+` is set instead of `+JENKINS_NODE_COOKIE+`, then parallel Pipeline stages running on the same agent may stop processes from other stages of the same build.
	Use `+JENKINS_NODE_COOKIE+` instead of `+BUILD_ID+` so that processes in parallel Pipeline stages are not unexpectedly stopped by the completion of other stages on that agent.

[jglick]

I do not recall offhand and I would not expect any text about this from the wiki to be accurate. Would need to be experimentally revalidated.

[MarkEWaite]

@LarrySul are you willing to do the experimental validation of my proposed text additions?

[bigdevlarry]

@MarkEWaite yes I'm open to doing the experimental validations

[MarkEWaite]

That's great @LarrySul. Thanks

Things to test include:

• Is JENKINS_NODE_COOKIE defined inside a Pipeline stage? (should confirm that it is)
• Are processes on the same agent that are started from different stages of the same job stopped when a single stage stops?
• Are processes on the same agent that are started from different stages of the same job stopped when the pipeline has completed and all stages are stopped?
• If JENKINS_NODE_COOKIE is defined with a different value in a stage, do long running processes started in that stage continue running after the stage is complete?
• If JENKINS_NODE_COOKIE is defined with a different value in a stage, do long running processes started in that stage continue running after the job is complete?
• If BUILD_ID is defined with a different value in a stage but JENKINS_NODE_COOKIE is unmodified, are long running processes from a stage stopped when the stage completes?
• If BUILD_ID is defined with a different value in a stage but JENKINS_NODE_COOKIE is unmodified, are long running processes from a stage stopped when the job completes?

[oleg-nenashev]

I think this is good to go. Sorry for the review delays

[kmartens27]

As this pull request has not had any updates since July 2021, and has been marked stalled (as discussed in Jenkins Documentation SIG office hours)[https://docs.google.com/document/d/1ygRZnVtoIvuEKpwNeF_oVRVCV5NKcZD1_HMtWlUZguo/edit#heading=h.cao5dku7jwfn], it will be closed and linked to the original issue that will give others an opportunity to continue the work already started here.