Skip to content

Runner reconciliation - #2556

Draft
cquil11 wants to merge 2 commits into
mainfrom
agent/remove-retired-b300-cw-runners
Draft

Runner reconciliation#2556
cquil11 wants to merge 2 commits into
mainfrom
agent/remove-retired-b300-cw-runners

Conversation

@cquil11

@cquil11 cquil11 commented Aug 11, 2026

Copy link
Copy Markdown
Collaborator

What changed

  • Reconciled Slurm runner mappings with the currently schedulable node capacity on each accessible cluster.
  • Standardized runner names as contiguous, zero-padded sequences beginning at _00.
  • Added the complete H100, B200-Nscale, GB200, GB300, H200-NB, and MI355X mappings.
  • Removed excess MI325X runner 09 and the retired B300 CoreWeave mapping.
  • Added cluster:b200-nscale hardware metadata and cluster mapping.

Operational changes

The corresponding GitHub runner registrations were configured directly on the cluster login nodes. Newly added and renamed listeners were intentionally left stopped/offline pending the node-lease rollout. Retired and excess registrations were removed. The two MI300X TensorWave runners were given the docker label.

Validation

  • Verified every reconciled cluster mapping is a contiguous two-digit sequence beginning at _00.
  • Parsed configs/runners.yaml successfully.
  • Ran git diff --check.
  • Ran uv run pytest -q utils/matrix_logic/test_validation.py: 126 passed.

@cquil11 cquil11 changed the title Remove retired B300 CoreWeave runners Runner reconciliation Aug 11, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

Status: No status

Development

Successfully merging this pull request may close these issues.

1 participant