Reword glossary entry for cluster #48686
Conversation
Clarified definition, especially wrt to control plane running on cluster members, the distinction between worker and control-plane nodes, and separation of control plane and workload in production environments.
The committers listed above are authorized under a signed CLA. |
k8s-ci-robot
added
the
cncf-cla: no
|
Welcome @TransitivityPaul! |
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
k8s-ci-robot
added
size/S
✅ Pull request preview available for checkingBuilt without sensitive environment variables
To edit notification comments on pull requests, go to your Netlify site configuration. |
| nodes and the Pods in the cluster. In production environments, the control plane usually | ||
| runs across multiple computers and a cluster usually runs multiple nodes, providing | ||
| fault-tolerance and high availability. | ||
| Machines may be physical or virtual. Machines that host pods are referred to as {{< glossary_tooltip text="nodes" term_id="node" >}}. Nodes that host workloads are referred to as worker nodes. There must be at least one worker node within a cluster. |
There was a problem hiding this comment.
No, this isn't correct. A cluster with zero worker nodes isn't very useful, but it's still technically a cluster (maybe you are auto scaling).
There was a problem hiding this comment.
Ah - OK. I kept this as it was part of the previous definition :)
| fault-tolerance and high availability. | ||
| Machines may be physical or virtual. Machines that host pods are referred to as {{< glossary_tooltip text="nodes" term_id="node" >}}. Nodes that host workloads are referred to as worker nodes. There must be at least one worker node within a cluster. | ||
|
|
||
| Control plane components run on one or more of the machines within the cluster. They may execute directly in the context of the operating system of a machine, or within pods. When control plane components run within pods on a machine, the machine may also be referred to as a control plane node. |
There was a problem hiding this comment.
No, this isn't correct. You can run the control plane within the cluster, but we actually recommend not doing that for the very best resilience.
There was a problem hiding this comment.
OK - All of the architecture diagrams show the control plane as being a part of the cluster. I don't mind either way as long as we're consistent. :). This is profoundly important from an architectural perspective. Is this the right place for that discussion?
There was a problem hiding this comment.
Control plane:
- always part of the cluster
- optional to run it on nodes
- if you run it on a separate server, the thing you run on a separate server is (if healthy) still logically a control plane, and that logical thing is part of the cluster
There was a problem hiding this comment.
Got it. I could co-host my control plane instance, with a bunch of other control planes for other clusters, on the same machines. I could host my control plane within another cluster that has it's own control plane.... etc. In tuo capite fiat! Will nudge this and other PRs to reflect. Thx.
There was a problem hiding this comment.
Please do update this when you have time to.
|
|
||
| Control plane components run on one or more of the machines within the cluster. They may execute directly in the context of the operating system of a machine, or within pods. When control plane components run within pods on a machine, the machine may also be referred to as a control plane node. | ||
|
|
||
| In production environments, the cluster typically consists of multiple machines, with the control plane components and workloads distributed across them for scale, resilience and high availability. It is a best practice in production environments to have workloads hosted on machines that do not host control plane components. |
There was a problem hiding this comment.
We don't formally recommend this topology; we think that a control plane that runs separately from the cluster is an even more resilient choice. However it is up to the cluster operator to make that call, not this project.
It is a best practice in production environments to have workloads hosted on machines that do not host control plane components.
True, but doesn't belong in the glossary. Instead, hyperlink to the page that makes the recommendation.
| aka: | ||
| tags: | ||
| - fundamental | ||
| - operation | ||
| --- | ||
| A set of worker machines, called {{< glossary_tooltip text="nodes" term_id="node" >}}, | ||
| that run containerized applications. Every cluster has at least one worker node. | ||
| A set of machines hosting containerized {{< glossary_tooltip text="workloads" term_id="workload" >}}, in the form of {{< glossary_tooltip text="pods" term_id="pod" >}}, and the {{< glossary_tooltip text="control plane" term_id="control-plane" >}} that manages them. |
There was a problem hiding this comment.
I think we can do better. Try this:
| A set of machines hosting containerized {{< glossary_tooltip text="workloads" term_id="workload" >}}, in the form of {{< glossary_tooltip text="pods" term_id="pod" >}}, and the {{< glossary_tooltip text="control plane" term_id="control-plane" >}} that manages them. | |
| A {{< glossary_tooltip text="control plane" term_id="control-plane" >}} and a set of worker machines that the | |
| control plane manages. A Kubernetes cluster lets you run containerized {{< glossary_tooltip text="workloads" term_id="workload" >}} | |
| as {{< glossary_tooltip text="pods" term_id="pod" >}}. |
There was a problem hiding this comment.
This depends on the other comment...
There was a problem hiding this comment.
Line 14 seems to be missing something?
There was a problem hiding this comment.
I revised. BTW you can accept suggestions as is, but you can also incorporate the spirit of the feedback and apply your own edits.
|
/retitle Reword glossary entry for cluster |
k8s-ci-robot
changed the title
|
/sig architecture |
k8s-ci-robot
added
the
sig/architecture
Clarified definition, especially wrt to control plane running on cluster members, the distinction between worker and control-plane nodes, and separation of control plane and workload in production environments.
Description
Issue
Closes: #