Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.
C++
Docker: an NVIDIA card the container cannot use ends the start before the model is prepared, with the --gpus all hint
Niko122183ae4131762f
diff --git a/docker/entrypoint.sh b/docker/entrypoint.shindex 13d652b22..edcfeff96 100755--- a/docker/entrypoint.sh+++ b/docker/entrypoint.sh@@ -137,9 +137,11 @@ MSGfi# the card is found through sysfs even when no device file was passed in: say so now instead of failing when the engine starts+# (before the model is prepared; /dev/dxg is the GPU of Docker Desktop on WSL2)backend=$(printf '%s' "$engine_json" | sed -n 's/.*"backend": *"\([a-z0-9]*\)".*/\1/p' | head -1)dev="${STRATA_DEV_DIR:-/dev}"case "$backend" in+ cuda) ls "$dev"/nvidia[0-9]* >/dev/null 2>&1 || [ -e "$dev/dxg" ] || { echo "[X] This computer has an NVIDIA card, but the container cannot use it, so nothing would run (and the model would be prepared for nothing). Start the container with: --gpus all (needs the NVIDIA Container Toolkit on the host; docs/DOCKER.md, Troubleshooting)." >&2; exit 3; } ;;hip) [ -e "$dev/kfd" ] || { echo "[X] This is an AMD card, but /dev/kfd is not in the container. Add: --device /dev/kfd --device /dev/dri --group-add video --group-add <render group number> (docs/DOCKER.md)" >&2; exit 3; } ;;sycl) ls "$dev"/dri/renderD* >/dev/null 2>&1 || { echo "[X] This is an Intel card, but /dev/dri is not in the container. Add: --device /dev/dri --group-add <render group number> (docs/DOCKER.md)" >&2; exit 3; } ;;esacdiff --git a/tools/test_docker_entrypoint.py b/tools/test_docker_entrypoint.pyindex 0a4f2eeb3..8ac8c4b1f 100644--- a/tools/test_docker_entrypoint.py+++ b/tools/test_docker_entrypoint.py@@ -223,6 +223,19 @@ class Entrypoint(unittest.TestCase):self.assertIn("strata-a.json", out)self.assertIn("does not have", out)+ def test_nvidia_card_without_gpu_access_ends_before_any_work(self):+ dev = self.tmp / "dev"+ dev.mkdir()+ r = subprocess.run([SH, str(self.script)], env=self.env(STUB_BACKEND="cuda", STRATA_DEV_DIR=str(dev), MODEL="Q2_0"),+ capture_output=True, text=True, timeout=30)+ self.assertEqual(r.returncode, 3)+ self.assertIn("--gpus all", r.stderr)+ self.assertNotIn("launch --no-tui", self.log.read_text())+ (dev / "nvidia0").write_text("") # with the device files the start goes on+ p = self.start(STUB_BACKEND="cuda", STRATA_DEV_DIR=str(dev), MODEL="Q2_0")+ self.wait_for("launch --no-tui")+ self.stop(p)+def test_other_commands_run_instead_of_the_server(self):r = subprocess.run([SH, str(self.script), "strata", "launch", "--plan", "--model", "q2_0"], env=self.env(), capture_output=True,text=True, timeout=30)