Skip to content

fix: expose accurate managed-server readiness - #3197

Merged
ananthsub merged 4 commits into
NVIDIA-NeMo:mainfrom
aditya-malte:fix/evaluator-health-readiness
Sep 10, 2026
Merged

ananthsub merged 4 commits into
NVIDIA-NeMo:mainfrom
aditya-malte:fix/evaluator-health-readiness

Conversation

@aditya-malte

Copy link
Copy Markdown
Contributor

What does this PR do?

  • Exposes a /health endpoint alongside the existing root liveness endpoint.
  • Keeps the managed head server at HTTP 503 while child servers and model endpoints are starting.
  • Marks the head server ready only after both readiness phases complete, after which /health returns HTTP 200.

NeMo Evaluator polls /health before creating its adapter proxy. Without this endpoint, the readiness check receives HTTP 404 and eventually times out. The readiness gate also prevents clients from sending work after child ports are published but before those services are actually ready.

Testing

  • 106 tests passed: tests/unit_tests/test_server_utils.py and tests/unit_tests/test_cli.py
  • Ruff check and format check passed for all changed files
  • pre-commit run --all-files passed

Checklist

  • I have read the contributing guidelines.
  • The change is focused; no unrelated edits are included.
  • Tests were added and pass locally.
  • Pre-commit checks pass locally.
  • All commits have DCO sign-off.

Signed-off-by: Aditya Uday Malte <aditya.malte@gmail.com>
Signed-off-by: Aditya Uday Malte <aditya.malte@gmail.com>
@copy-pr-bot

copy-pr-bot Bot commented Sep 9, 2026

Copy link
Copy Markdown

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@nemo-automation-bot nemo-automation-bot Bot added the community-request Issue reported or requested by someone from the community label Sep 9, 2026

@ananthsub ananthsub left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

thanks for the contribution! i added 2 comments inline and then this should be good to merge

Comment thread nemo_gym/server_utils.py
Comment thread nemo_gym/server_utils.py
@yaoyu-33 yaoyu-33 added area:core Shared APIs, servers, telemetry, health, and registries bug Something isn't working waiting-on-customer Waiting on the original author to respond labels Sep 9, 2026
@ananthsub

Copy link
Copy Markdown
Contributor

/ok to test 4f229ac

@ananthsub
ananthsub merged commit d33e045 into NVIDIA-NeMo:main Sep 10, 2026
38 checks passed
@svcnvidia-nemo-ci svcnvidia-nemo-ci removed the waiting-on-customer Waiting on the original author to respond label Sep 11, 2026
jkyi-nvidia pushed a commit that referenced this pull request Sep 11, 2026
## What does this PR do?

- Exposes a /health endpoint alongside the existing root liveness
endpoint.
- Keeps the managed head server at HTTP 503 while child servers and
model endpoints are starting.
- Marks the head server ready only after both readiness phases complete,
after which /health returns HTTP 200.

NeMo Evaluator polls /health before creating its adapter proxy. Without
this endpoint, the readiness check receives HTTP 404 and eventually
times out. The readiness gate also prevents clients from sending work
after child ports are published but before those services are actually
ready.

## Testing

- 106 tests passed: tests/unit_tests/test_server_utils.py and
tests/unit_tests/test_cli.py
- Ruff check and format check passed for all changed files
- pre-commit run --all-files passed

## Checklist

- [x] I have read the contributing guidelines.
- [x] The change is focused; no unrelated edits are included.
- [x] Tests were added and pass locally.
- [x] Pre-commit checks pass locally.
- [x] All commits have DCO sign-off.

---------

Signed-off-by: Aditya Uday Malte <aditya.malte@gmail.com>
Co-authored-by: audaymalte-web <audaymalte@nvidia.com>
Signed-off-by: Junkeun Yi <jkyi@nvidia.com>

This branch was successfully deployed

1 active deployment
public 4f229acb Deployed Sep 10, 2026 by copy-pr-bot[bot] via release / finalize / notify #2775
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

area:core Shared APIs, servers, telemetry, health, and registries bug Something isn't working community-request Issue reported or requested by someone from the community

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants