Skip to content

[iris] Label always-on CoreWeave nodes as system-critical#4011

Merged
rjpower merged 1 commit intomainfrom
coreweave-system-critical-label
Mar 23, 2026
Merged

[iris] Label always-on CoreWeave nodes as system-critical#4011
rjpower merged 1 commit intomainfrom
coreweave-system-critical-label

Conversation

@rjpower
Copy link
Copy Markdown
Collaborator

@rjpower rjpower commented Mar 23, 2026

Add cks.coreweave.cloud/system-critical label to NodePools with min_nodes > 0.
This pins Konnectivity agents and monitoring pods to always-on CPU nodes so
GPU NodePools can safely scale to zero without losing cluster connectivity.

@rjpower rjpower added the agent-generated Created by automation/agent label Mar 23, 2026
@claude
Copy link
Copy Markdown
Contributor

claude Bot commented Mar 23, 2026

Claude finished @rjpower's task in 37s —— View job


PR Review

  • Read changed file and diff
  • Analyze correctness
  • Post review

This PR conditionally adds the cks.coreweave.cloud/system-critical: "true" label to NodePools with min_nodes > 0, ensuring Konnectivity agents and monitoring pods land on always-on nodes. The change is correct and matches the stated objective.

No issues found. The conditional via dict unpacking is clean, the condition (min_nodes > 0) correctly identifies always-on pools, and the label is only applied where it makes sense.

LGTM.

@rjpower rjpower merged commit a20b0b5 into main Mar 23, 2026
41 checks passed
@rjpower rjpower deleted the coreweave-system-critical-label branch March 23, 2026 20:19
Helw150 pushed a commit that referenced this pull request Apr 8, 2026
Add cks.coreweave.cloud/system-critical label to NodePools with
min_nodes > 0.
This pins Konnectivity agents and monitoring pods to always-on CPU nodes
so
GPU NodePools can safely scale to zero without losing cluster
connectivity.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

agent-generated Created by automation/agent

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant