Merge pull request 'fix: retry NvAPI init on autoload to handle early boot race' (#3) from fix/autoload-nvapi-retry into main

Reviewed-on: #3

nvcurve/hal/gpu.py

@@ -12,8 +12,7 @@ def init_nvapi() -> None:
     """Initialize NvAPI. Must be called before any GPU operations."""
     init_fn = query_interface(FUNC["Initialize"], nargs=0)
     if not init_fn or init_fn() != 0:
-        print("NvAPI_Initialize failed")
+        raise RuntimeError("NvAPI_Initialize failed")
-        sys.exit(1)
 
 
 def enumerate_gpus() -> tuple[ctypes.Array, int]:

nvcurve/profiles/apply.py

@@ -162,10 +162,18 @@ def run_autoload() -> None:
     from ..hal.gpu import init_nvapi, discover_gpus
     from ..hal.monitoring import init_nvml, shutdown_nvml
 
+    # Retry NvAPI init — the driver may not be fully ready at early boot.
+    nvapi_ready = False
+    for attempt in range(1, 17):
         try:
             init_nvapi()
+            nvapi_ready = True
+            break
         except Exception as exc:
-        log.error("Failed to initialize NvAPI: %s", exc)
+            log.warning("NvAPI init attempt %d/16 failed: %s", attempt, exc)
+            time.sleep(2)
+    if not nvapi_ready:
+        log.error("Failed to initialize NvAPI after 16 attempts.")
         sys.exit(1)
 
     init_nvml()  # best-effort