SolitaryThinker 07f1cfc356 ray executor: wait out async GPU release on shutdown; retry node GPU probe on init ...

killed actors free their devices asynchronously — a load fired right after a
release could still see 0 GPUs on the node and hard-fail. shutdown now waits
(<=60s) for the GPUs to return to ray's ledger, and initialize_ray_cluster's
node-local GPU probe retries for 60s instead of failing instantly.

2026-08-07 00:05:38 -07:00

..

__init__.py

[Feat] add ray support (#838)

2025-10-16 23:17:54 -07:00

executor.py

[feat] LoRA controls and integration for Dreamverse (#1420)

2026-06-01 18:54:11 -07:00

gpu_worker.py

[bugfix]: skip unused output materialization (#1567)

2026-07-16 04:03:07 +00:00

multiproc_executor.py

[model] Flux2 Klein Port (#1349)

2026-06-09 14:55:55 -07:00

ray_distributed_executor.py

ray executor: wait out async GPU release on shutdown; retry node GPU probe on init

2026-08-07 00:05:38 -07:00

ray_env.py

[feat] pre-commit support 120 col num (#1167)

2026-03-11 20:19:30 -07:00

ray_utils.py

ray executor: wait out async GPU release on shutdown; retry node GPU probe on init

2026-08-07 00:05:38 -07:00

worker_base.py

[feat] pre-commit support 120 col num (#1167)

2026-03-11 20:19:30 -07:00