fix/nvidia-gpu-load #10

Merged
litelotus merged 4 commits from fix/nvidia-gpu-load into main 2026-08-18 21:44:48 +00:00
Owner
No description provided.
- gpu-nvidia module: boot.kernelModules = [nvidia nvidia_uvm] — headless
  hosts have no display stack to trigger module loading, so nvidia-smi
  fails with 'couldn't communicate with the NVIDIA driver'
- 01-server: pin nvidiaPackages.legacy_580 — Tesla M10 (GM107GL) is not
  on the 595.x supported-products list; 580 is the last ESR branch for
  Maxwell
feat(hermes): add firecrawl dependency group
Some checks failed
ci/crow/push/check Pipeline failed
18e05414ff
web_search/web_extract use the firecrawl-py client, which was missing
from the sealed hermes venv (no pip/ensurepip in Nix env, so the
auto-install fallback failed). firecrawl is a declared optional
dependency group in hermes-agent's pyproject.toml, so it slots into
extraDependencyGroups alongside honcho.
feat(hermes): add voice + vision dependency groups
Some checks failed
ci/crow/push/check Pipeline failed
55e070df31
voice: local STT is enabled (stt.provider: local) but faster-whisper
and sounddevice were missing from the sealed venv — same latent
failure as firecrawl.
vision: vision_tools lazy-imports Pillow, which was absent.
feat(odin): add tea (Gitea/Forgejo CLI)
Some checks failed
ci/crow/push/check Pipeline failed
81740598ff
tea 0.14.0 from nixpkgs — the official Gitea CLI, API-compatible
with the homelab Forgejo instance. Enables PR/issue/repo management
from the terminal via `tea login`.
Sign in to join this conversation.
No reviewers
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
litelotus/.nix!10
No description provided.