Remove B200 GPU module (reservation expired, not renewed) - #975
Open
jeanschmidt wants to merge 1 commit into
Open
Remove B200 GPU module (reservation expired, not renewed)#975jeanschmidt wants to merge 1 commit into
jeanschmidt wants to merge 1 commit into
Conversation
**Impact:** OSDC prod (us-east-2 loses B200 runners); PyTorch B200 CI jobs re-route to the existing DGX runners **Risk:** medium ## What Deletes the `nodepools-b200` and `arc-runners-b200` delegate modules, their defs and node-setup script, drops the B200 capacity reservation and module wiring from `clusters.yaml`, and scrubs all B200 references from docs, skills, tests, and shared Python (instance specs, label maps, fleet naming). ## Why The AWS Capacity Block reservation backing the B200 pool (`cr-0b15c0b3163f09d26`, us-east-2) expired and we've decided not to renew it. With no reserved capacity there's nothing for the module to schedule onto, so it's removed rather than left dangling. # Notes - Routing change (`pytorch-runners-routing` skill + `.github/arc.yaml` in the pytorch repo): `linux.dgx.b200` moves from a `meta_only_runners` override back to an **identity passthrough**, so both `linux.dgx.b200` and `linux.dgx.b200.8` now land on the existing DGX runners instead of OSDC ARC. The actual `arc.yaml` / `map_ec2_to_arc.py` edits live in `pytorch/pytorch` and are not part of this diff — only the OSDC-side skill docs are updated here. - us-west-1 was H100-only already; no B200 capacity there, so no functional change beyond doc cleanup. - Also cleans up leftover git conflict markers in `docs/observability-estimates.md`. - After merge, `just lint` / `just test` should be re-run to confirm the def/script/Python sync is complete. Signed-off-by: Jean Schmidt <contato@jschmidt.me>
jeanschmidt
temporarily deployed
to
osdc-staging
August 4, 2026 21:05 — with
GitHub Actions
Inactive
Capacity reportcommit ✅ simulate-cluster✅ analyze-utilization |
tofu plan — meta-prod-aws-ue2✅ Plan succeeded · commit Plan output |
jeanschmidt
temporarily deployed
to
osdc-staging
August 4, 2026 21:06 — with
GitHub Actions
Inactive
tofu plan — meta-prod-aws-uw1✅ Plan succeeded · commit Plan output |
jeanschmidt
temporarily deployed
to
osdc-staging
August 4, 2026 21:07 — with
GitHub Actions
Inactive
tofu plan — meta-prod-aws-ue1✅ Plan succeeded · commit Plan output |
jeanschmidt
temporarily deployed
to
osdc-staging
August 4, 2026 21:08 — with
GitHub Actions
Inactive
tofu plan — lf-prod-aws-ue1✅ Plan succeeded · commit Plan output |
jeanschmidt
temporarily deployed
to
osdc-staging
August 4, 2026 21:09 — with
GitHub Actions
Inactive
tofu plan — lf-prod-aws-ue2✅ Plan succeeded · commit Plan output |
huydhn
approved these changes
Aug 4, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Impact: OSDC prod (us-east-2 loses B200 runners)
Risk: low
What
Deletes the
nodepools-b200andarc-runners-b200delegate modules, their defs and node-setup script, drops the B200 capacity reservation and module wiring fromclusters.yaml, and scrubs all B200 references from docs, skills, tests, and shared Python (instance specs, label maps, fleet naming).Why
The AWS Capacity Block reservation backing the B200 pool (
cr-0b15c0b3163f09d26, us-east-2) expired and we've decided not to renew it. With no reserved capacity there's nothing for the module to schedule onto, so it's removed rather than left dangling.Notes
pytorch-runners-routingskill +.github/arc.yamlin the pytorch repo):linux.dgx.b200moves from ameta_only_runnersoverride back to an identity passthrough, so bothlinux.dgx.b200andlinux.dgx.b200.8now land on the existing DGX runners instead of OSDC ARC. The actualarc.yaml/map_ec2_to_arc.pyedits live inpytorch/pytorchand are not part of this diff — only the OSDC-side skill docs are updated here.docs/observability-estimates.md.just lint/just testshould be re-run to confirm the def/script/Python sync is complete.Playbook for removal: