Hi all,

We made decisions on a few terms at the yesterday's governance meeting:

   - Master node => "Built-in Node"
   - "master" label => "built-in" // We will use it unless we discover a 
   technical issue with the hyphen. Then we fallback to “builtin” 
   - “Master branch” in documentation and help => "default branch"
   - Agent-to-Master security => " Agent-to-Controller security "
   - "Jenkins master container " => "Jenkins controller container"
   - "Serialization whitelist" for JEP-200 => "serialization allowlist"

We also agreed that we will be using "allowlist" in our terminology, not 
the "permitlist" as it was suggested in a few occasions. We have not 
finalized decisions on other terms, including the "Jenkins master pod". I 
raised https://github.com/jenkinsci/kubernetes-operator/issues/561 in the 
Jenkins operator project to track the change on its side once we agree on 
the term.

If anyone is interested, I can create a global "terminology cleanup" 
project in the jenkinsci organization. It will allow tracking pull request 
better on the GitHub's side

Best regards,
Oleg Nenashev


On Wednesday, May 5, 2021 at 12:02:42 PM UTC+2 Daniel Beck wrote:

>
>
> > On 4. May 2021, at 16:59, Oleg Nenashev <[email protected]> wrote:
> > 
> > • Master node => "Built-in Node"
>
> To provide a bit of context for this one for those that don't remember 
> from last year :-)
>
> Before, there was no real distinction between "Jenkins master, the 
> process" (mostly) and "Jenkins master, the node". When I worked on the PR 
> in which I started cleaning up the terms, it became apparent a different 
> term could be useful.[1]
>
> A simple example: The built-in node can be offline while the controller is 
> otherwise running.
>
> In some code, the relation between master-specific and global node 
> properties also wasn't clear in some places because both were occasionally 
> called "master" (and only one set is inherited by agents).
>
> There's not a huge list of obvious examples because a lot of the things 
> that could matter are shared (process, file system, config file to an 
> extent) or irrelevant (node launcher).
>
> I still think it would be useful to distinguish in terms between the 
> controller and the built-in node, if only because 'controller' for the node 
> may create wrong associations (it controlling things, rather than "just" 
> being part of the controller process).
>
>
> However there are also limitations, which make a different term not an 
> obviously correct choice:
>
> - The built-in node is part of the controller process, it shares the 
> controller's file system and OS permissions. If the built-in node is doing 
> work, the controller has load. A lot of resources are shared, so "the 
> built-in node's configuration is stored in the config.xml file with most of 
> the controller configuration on the controller file system" etc.
> - People seem to confuse executors and nodes/agents fairly regularly, so 
> may well consider these to be the same thing because the differences are 
> way less relevant than compared to agents, leading to wrong documentation 
> and other advice, possibly confusing those aware of the terms. (It might 
> help that controller as a term is getting rather well established, and that 
> the node will get labels (both UI and environment var) referring to it by 
> its new name, but who knows.) 
>
>
> I encourage you to check out the PR with placeholder term to get a sense 
> for the differences and consider whether you think distinguishing the terms 
> is useful. As the PR is still a draft and uses an obvious placeholder term, 
> please skip doing an actual review for now.
>
> (Note that the behavior-changing code in my PR (related to migration) 
> would be needed anyway, regardless of the term we choose. It's more about 
> removing "master" than what the replacement term is.)
>
>
> 1: https://github.com/jenkinsci/jenkins/pull/5425
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"Jenkins Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/jenkinsci-dev/8507bd02-fbad-4844-aeb9-b1ee58753326n%40googlegroups.com.

Reply via email to