spark recipe: entrypoint (clean image is ENTRYPOINT []) + MAX_JOBS=2 (JIT nvcc OOM at boot); clean r12 image serving, exact-391 verified

recipes/glm53-flash-nvfp4-spark/recipe.toml

@@ -70,6 +70,11 @@ engine = "vllm"
 name = "GLM-5.3-Flash NVFP4-Spark (vLLM, 2x Spark, MTP-5)"
 docker_image = "glm53-jovian-r12:local"  # r12 fork build (see docker-images/glm53-jovian-r12)
 docker_image_dir = "glm53-jovian-r12"
+# Required: the clean r12 image follows the solar contract (ENTRYPOINT []
+# + CMD bash), so the recipe carries the entrypoint (the old day-0-based
+# r12 image had ["vllm","serve"] baked in, which is why this was absent
+# until the clean rebuild).
+entrypoint = ["vllm", "serve"]
 model = "local-inference-lab/GLM-5.3-Flash-NVFP4-Spark"
 required_nodes = 2
 tensor_parallel_size = 2
@@ -136,6 +141,11 @@ args = [
 
 [env]
 HF_HUB_OFFLINE = "1"
+# FlashInfer JIT (mxfp8_gemm_cutlass_sm120, ~12 kernels) compiles at first
+# boot with 12 parallel nvcc jobs -> OOM-killed (code 137) on GB10's
+# 121 GB unified while the 93 GB weights + 10 GB KV pin are resident.
+# MAX_JOBS caps ninja parallelism so the JIT fits in the remaining ~20 GB.
+MAX_JOBS = "2"
 # NoPE-MLA rope-pad runtime gate (mod baked into the image)
 VLLM_MLA_NOPE_PAD_ROPE = "1"
 VLLM_ENGINE_READY_TIMEOUT_S = "3600"