Merge branch 'main' into mcv-binary-cache

Author: maryamtahhan

.pre-commit-config.yaml

@@ -19,12 +19,12 @@ repos:
         additional_dependencies:
           - tomli
   - repo: https://github.com/alessandrojcm/commitlint-pre-commit-hook
-    rev: v9.23.0
+    rev: v9.24.0
     hooks:
       - id: commitlint
         stages: [commit-msg]
   - repo: https://github.com/adrienverge/yamllint
-    rev: v1.37.1
+    rev: v1.38.0
     hooks:
       - id: yamllint
 exclude: ^mcv/vendor/.*|^mcv/_output/.*|^venv/.*|^mcv/logo/.*|^mcv/example/vllm-cache/.*|^mcv/example/.*-binary-cache/.*

mcm/model_cache_manager/data/kernel_validator.py

@@ -16,15 +16,15 @@
 log = logging.getLogger(__name__)
 
 
-class KernelTarget(BaseModel):
+class KernelTarget(BaseModel):  # pylint: disable=too-few-public-methods
     """Validation model for the 'target' field of kernel metadata."""
 
     backend: str = ""
     arch: int | str = 0
     warp_size: int = 0
 
 
-class KernelMetadata(BaseModel):
+class KernelMetadata(BaseModel):  # pylint: disable=too-few-public-methods
     """
     Pydantic model for validating kernel metadata.
 

mcm/model_cache_manager/tests/utils/test_cache_mode_detection.py

@@ -244,6 +244,23 @@ def test_detect_vllm_structure_no_rank_dirs(self):
         mode = detect_cache_mode(cache_dir)
         self.assertEqual(mode, MODE_TRITON)
 
+    def test_detect_vllm_binary_cache_structure(self):
+        """Test detection of new vLLM cache structure with binary artifacts."""
+        # Binary artifacts are files (not directories) named artifact_compile_range_*
+        vllm_dir = self.temp_dir / "vllm_binary"
+        torch_compile = vllm_dir / "torch_compile_cache"
+        hash_dir = torch_compile / "some_vllm_hash"
+        rank_dir = hash_dir / "rank_0_0"
+        backbone_dir = rank_dir / "backbone"
+        backbone_dir.mkdir(parents=True)
+
+        # Create a binary artifact file (not a directory)
+        binary_artifact = backbone_dir / "artifact_compile_range_0"
+        binary_artifact.write_bytes(b"\x00binary data")
+
+        mode = detect_cache_mode(vllm_dir)
+        self.assertEqual(mode, MODE_VLLM)
+
     def test_detect_mixed_legacy_and_triton_vllm_takes_precedence(self):
         """Test detection when both vllm-legacy and triton structures exist."""
         cache_dir = self.temp_dir / "mixed_cache"

mcm/model_cache_manager/utils/utils.py

@@ -198,8 +198,12 @@ def iter_artifact_compile_range_dirs(backbone_dir: Path):
 
 
 def _has_artifact_compile_range_with_triton(backbone_dir: Path) -> bool:
-    """Check if backbone directory has artifact_compile_range subdirectories with triton."""
+    """Check if backbone directory has artifact_compile_range artifacts (unpacked or binary)."""
     for artifact_dir in iter_artifact_compile_range_dirs(backbone_dir):
+        # Binary artifact (file) — contains packed triton data
+        if artifact_dir.is_file():
+            return True
+        # Unpacked artifact (directory) — check for triton subdirectory
         triton_dir = artifact_dir / "triton"
         if triton_dir.exists():
             return True

mcm/pyproject.toml

@@ -10,18 +10,21 @@ authors         = [{name = "Alessandro Sangiorgi", email = "asangior@redhat.com"
 requires-python = ">=3.9"
 
 dependencies = [
-  "triton",
   "rich>=13.7",
   "typer[all]",
   "structlog",
   "pydantic>=2",
   "pydantic-settings>=2",
   "sqlalchemy>=2.0.40",
+  "vllm==0.15.1",
 ]
 
 [project.scripts]
 mcm = "model_cache_manager.cli.main:run"
 
+[tool.pylint."messages_control"]
+disable = ["import-error"]
+
 [tool.setuptools]
 package-dir = {"" = "."}
 

mcm/requirements.txt

@@ -1,8 +1,8 @@
-typer>=0.9.0
+typer[all]
 rich>=13.7
 pydantic>=2.0.0
 pydantic-settings>=2.0.0
-structlog>=23.0.0
+structlog
 sqlalchemy>=2.0.40
 pytest>=7.0.0
-vllm
+vllm==0.15.1

mcv/go.mod

@@ -7,7 +7,7 @@ require (
 	github.com/containers/buildah v1.42.1
 	github.com/containers/podman/v5 v5.7.0
 	github.com/docker/docker v28.5.2+incompatible
-	github.com/google/go-containerregistry v0.20.6
+	github.com/google/go-containerregistry v0.20.7
 	github.com/jaypipes/ghw v0.17.0
 	github.com/jaypipes/pcidb v1.0.1
 	github.com/pkg/errors v0.9.1
@@ -52,7 +52,7 @@ require (
 	github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc // indirect
 	github.com/disiqueira/gotree/v3 v3.0.2 // indirect
 	github.com/distribution/reference v0.6.0 // indirect
-	github.com/docker/cli v28.5.2+incompatible // indirect
+	github.com/docker/cli v29.0.3+incompatible // indirect
 	github.com/docker/distribution v2.8.3+incompatible // indirect
 	github.com/docker/docker-credential-helpers v0.9.4 // indirect
 	github.com/docker/go-connections v0.6.0 // indirect
@@ -114,9 +114,9 @@ require (
 	github.com/proglottis/gpgme v0.1.5 // indirect
 	github.com/seccomp/libseccomp-golang v0.11.1 // indirect
 	github.com/secure-systems-lab/go-securesystemslib v0.9.1 // indirect
-	github.com/sigstore/fulcio v1.8.3 // indirect
+	github.com/sigstore/fulcio v1.8.5 // indirect
 	github.com/sigstore/protobuf-specs v0.5.0 // indirect
-	github.com/sigstore/sigstore v1.10.0 // indirect
+	github.com/sigstore/sigstore v1.10.3 // indirect
 	github.com/skeema/knownhosts v1.3.2 // indirect
 	github.com/smallstep/pkcs7 v0.1.1 // indirect
 	github.com/spf13/pflag v1.0.10 // indirect
@@ -135,17 +135,18 @@ require (
 	go.opentelemetry.io/otel/metric v1.38.0 // indirect
 	go.opentelemetry.io/otel/trace v1.38.0 // indirect
 	go.yaml.in/yaml/v2 v2.4.3 // indirect
-	golang.org/x/crypto v0.45.0 // indirect
-	golang.org/x/mod v0.29.0 // indirect
-	golang.org/x/net v0.47.0 // indirect
-	golang.org/x/sync v0.18.0 // indirect
-	golang.org/x/sys v0.38.0 // indirect
-	golang.org/x/term v0.37.0 // indirect
-	golang.org/x/text v0.31.0 // indirect
-	google.golang.org/genproto/googleapis/api v0.0.0-20251022142026-3a174f9686a8 // indirect
-	google.golang.org/genproto/googleapis/rpc v0.0.0-20251103181224-f26f9409b101 // indirect
-	google.golang.org/grpc v1.77.0 // indirect
-	google.golang.org/protobuf v1.36.10 // indirect
+	golang.org/x/crypto v0.46.0 // indirect
+	golang.org/x/mod v0.31.0 // indirect
+	golang.org/x/net v0.48.0 // indirect
+	golang.org/x/sync v0.19.0 // indirect
+	golang.org/x/sys v0.39.0 // indirect
+	golang.org/x/term v0.38.0 // indirect
+	golang.org/x/text v0.32.0 // indirect
+	golang.org/x/tools v0.40.0 // indirect
+	google.golang.org/genproto/googleapis/api v0.0.0-20251222181119-0a764e51fe1b // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20251222181119-0a764e51fe1b // indirect
+	google.golang.org/grpc v1.78.0 // indirect
+	google.golang.org/protobuf v1.36.11 // indirect
 	gopkg.in/tomb.v1 v1.0.0-20141024135613-dd632973f1e7 // indirect
 	gopkg.in/yaml.v3 v3.0.1 // indirect
 	howett.net/plist v1.0.0 // indirect