daily-fix: cpu intents in _DEFAULT_GPUS_FOR_INTENT
kind: infra#wf-fix#wf-fix-fp:c4b22d75da93#daily-auto-filed
track:Highlight text on any card (body, plan, or an event) to anchor a comment, or leave a whole-task comment here. Mention @claude to summon a reply.
No comments yet.
Overview / Motivation
Auto-filed by /daily 2026-07-16 from transcript mining (#1336 E1 cpu-mid dispatch walked every lane exit-2; the boot-disk-floor half of the same incident was filed as #1374 — this is the other half).
Goal
Make CPU-intent dispatches rank the free SLURM lanes instead of degrading to unranked.
Workflow gap
- Bug observed: slurm.py's _DEFAULT_GPUS_FOR_INTENT (L386) lacks the cpu-small/cpu-mid/cpu-bigmem intents, so the router's estimate path raises 'no default GPU count for intent cpu-mid' and the free SLURM lanes degrade to 'treated as unranked' on CPU dispatches (#1336 E1 dispatch, ~00:37Z)
- Why it is a workflow gap: The #747 CPU lanes are first-class; an estimate raise that silently unranks free lanes skews routing toward paid rungs.
- Confidence (emitter): medium
- verified-at-filing:
grep -n '_DEFAULT_GPUS_FOR_INTENT' src/explore_persona_space/backends/slurm.py-> dict at L386 (keys: lora-7b/lora/eval/debug/ft-7b/inf-70b/ft-70b; no cpu-* — absence claim; raise path at L409); incident text in #1336 transcript (~00:37Z)
Proposed change (candidate diff sketch — refine in planning)
add the cpu-* intents (0 GPUs) to _DEFAULT_GPUS_FOR_INTENT (or short-circuit the estimate for gpu_count=0 intents) so CPU dispatches rank the free lanes normally
Scope / surfaces
- Primary target:
src/explore_persona_space/backends/slurm.py - Grep the workflow surface for the pattern before editing and update every hit; list them in the plan.
Constraints / invariants
scripts/workflow_lint.py --check-askspasses; ruff on touched files passes.- This session runs under a
workflow_fix_target:Provenance line — it MUST NOT auto-route its own subagents' workflow-fix candidates (recursion guard).
Provenance
- fingerprint: c4b22d75da93
Activity