app/backend/fluksio at 53b49e5f683cbaccec0d68068f7c7414df96fbaf - app

Files

T

stroblmeandClaude Opus 5 53b49e5f68 Retire the GPU workers when a run that held a card finishes

A worker that has run a jax node keeps holding the GPU after the run: XLA
preallocates most of the VRAM at import and never releases it, so the next
process OOMs on preallocation while a warm idle worker sits on the card.
Pools are kept warm on purpose — a library reads its environment at import,
so a warm worker cannot be re-told — but the end of a run is a point where
the memory should go back, and the environments carrying a GPU assignment
are exactly the pools that ran on one. Idle ones go now, busy ones when
they return.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019Hra4ndWMCLU5F3KjUuVAc

2026-08-29 13:53:11 +02:00

alembic

Let a sweep run more than four at a time, and name the run a failure was in

2026-08-27 14:17:51 +02:00

api

Fetch a run's artifacts from the CLI

2026-08-29 13:48:23 +02:00

cloud

Style the engine's own logs, notice enrolment while serving, say more in status

2026-08-25 09:29:22 +02:00

core

Refuse a zero concurrency limit instead of reading it as the default