Skip to content

Add SBD V6 metadata and startup timeline - #1331

Draft
luochen-amd wants to merge 1 commit into
mainfrom
feat/sbd-v6-metadata-install-model-gate
Draft

Add SBD V6 metadata and startup timeline#1331
luochen-amd wants to merge 1 commit into
mainfrom
feat/sbd-v6-metadata-install-model-gate

Conversation

@luochen-amd

Copy link
Copy Markdown
Contributor

Summary

  • add the additive SBD V6 metadata projection while preserving the V5 schema and output
  • persist ordered install and model-gate timeline events for fresh and resumed runs
  • preserve hard preflight failures, including failures outside instrumented install steps
  • retain startup timeline history across resume and recover partial legacy migrations

Scope

  • metadata
  • outcome projection
  • install timeline
  • model-gate timeline
  • top-level V6 close placeholder

Framework Agent timeline collection is intentionally excluded and tracked on a separate branch.

Compatibility

  • keeps the existing V5 schema version and fields unchanged
  • treats V6 collection and persistence as best-effort telemetry
  • does not change preflight, model-gate, resume, or optimization outcomes

Validation

  • 21 focused SBD V6 and session-package tests passed
  • Ruff formatting and lint checks passed for all changed Python files
  • git diff check passed

Environment Note

Windows test runs used a no-op fcntl compatibility module for imports. Linux CI exercises the production locking implementation.



def test_preflight_hard_failure_creates_session_and_final_sbd(tmp_path, monkeypatch):
import hyperloom.inference_optimizer.cli as optimizer_cli


def test_unwrapped_preflight_failure_is_persisted_as_failed(tmp_path, monkeypatch):
import hyperloom.inference_optimizer.cli as optimizer_cli


def test_busy_resume_preflight_failure_uses_isolated_session(tmp_path, monkeypatch):
import hyperloom.inference_optimizer.cli as optimizer_cli


def test_preflight_hard_failure_creates_session_and_final_sbd(tmp_path, monkeypatch):
import hyperloom.inference_optimizer.cli as optimizer_cli


def test_unwrapped_preflight_failure_is_persisted_as_failed(tmp_path, monkeypatch):
import hyperloom.inference_optimizer.cli as optimizer_cli


def test_busy_resume_preflight_failure_uses_isolated_session(tmp_path, monkeypatch):
import hyperloom.inference_optimizer.cli as optimizer_cli
@luochen-amd
luochen-amd force-pushed the feat/sbd-v6-metadata-install-model-gate branch from f43bd36 to 937c896 Compare August 28, 2026 08:58
@luochen-amd
luochen-amd marked this pull request as draft August 28, 2026 09:48
@github-actions

Copy link
Copy Markdown

CI E2E report — ❌ Timeout

item value
result ❌ Timeout
model Qwen/Qwen3-0.6B (dense)
resources 1× GPU, TP=1
PR branch feat/sbd-v6-metadata-install-model-gate
commit 937c896fa1fec1162e9881543ef192d6a5470653
session_id 50d81c88-9c73-4668-a744-e86c0b16224c
queue → dispatch 24m 1s
run time 77m 16s
total 101m 17s
reason Timed out — the run never reached a terminal state in time (task stuck, or the GPU stayed queued too long).
detail not terminal after 13200s

details

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants