Files
wehub-resource-sync 94057c3d3e
PR Test (NPU) / check-changes (push) Has been cancelled
PR Test (NPU) / pr-gate (push) Has been cancelled
PR Test (NPU) / set-image-config (push) Has been cancelled
PR Test (NPU) / stage-b-test-1-npu-a2 (0) (push) Has been cancelled
PR Test (NPU) / stage-b-test-1-npu-a2 (1) (push) Has been cancelled
PR Test (NPU) / stage-b-test-2-npu-a2 (0) (push) Has been cancelled
PR Test (NPU) / stage-b-test-2-npu-a2 (1) (push) Has been cancelled
PR Test (NPU) / stage-b-test-4-npu-a3 (push) Has been cancelled
PR Test (NPU) / stage-b-test-16-npu-a3 (push) Has been cancelled
PR Test (NPU) / multimodal-gen-test-1-npu-a3 (push) Has been cancelled
PR Test (NPU) / multimodal-gen-test-2-npu-a3 (push) Has been cancelled
PR Test (Arm64) / pr-gate (push) Has been cancelled
PR Test (Arm64) / check-changes (push) Has been cancelled
PR Test (Arm64) / build-test (push) Has been cancelled
PR Test (sgl-router) / gate (push) Has been cancelled
PR Test (sgl-router) / tier-1 — lint (push) Has been cancelled
PR Test (sgl-router) / tier-2 — build + test (push) Has been cancelled
PR Test (sgl-router) / tier-3 — docker (placeholder) (push) Has been cancelled
PR Test (sgl-router) / tier-3 — k8s integration (push) Has been cancelled
PR Test (sgl-router) / tier-3 — e2e (push) Has been cancelled
PR Test (sgl-router) / finish (push) Has been cancelled
PR Test (NPU) / single-node-poc (map[name:qwen3_6_27b_w8a8_1p_in64k_out1k_50ms runner:linux-aarch64-a3-2 test_case:test/registered/ascend/performance/qwen3_6_27b/test_npu_qwen3_6_27b_w8a8_1p_in64k_out1k_50ms.py test_type:perf]) (push) Has been cancelled
PR Test (NPU) / pr-test-npu-finish (push) Has been cancelled
PR Test (Xeon) / pr-gate (push) Has been cancelled
PR Test (Xeon) / check-changes (push) Has been cancelled
PR Test (Xeon) / build-test (, xeon-gnr, base-b-test-cpu) (push) Has been cancelled
PR Test (XPU) / check-changes (push) Has been cancelled
PR Test (XPU) / pr-gate (push) Has been cancelled
PR Test (XPU) / stage-a-test-1-gpu-xpu (push) Has been cancelled
PR Test (XPU) / wait-for-stage-a (push) Has been cancelled
PR Test (XPU) / stage-b-test-1-gpu-xpu (push) Has been cancelled
PR Test (XPU) / finish (push) Has been cancelled
CI Model Inventory / build-inventory (push) Has been cancelled
Lint / lint (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark Compilation Check (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark - Manual Policy (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark - Request Processing (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark Summary (push) Has been cancelled
PR Test (SMG) / build-wheel (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on windows (x86_64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on macos (x86_64 - auto) (push) Has been cancelled
PR Test (SMG) / python-unit-tests (push) Has been cancelled
PR Test (SMG) / unit-tests (push) Has been cancelled
PR Test (SMG) / benchmarks (push) Has been cancelled
PR Test (SMG) / chat-completions (push) Has been cancelled
PR Test (SMG) / chat-completions-4gpu (push) Has been cancelled
PR Test (SMG) / e2e (push) Has been cancelled
PR Test (SMG) / docker-build-test (push) Has been cancelled
PR Test (SMG) / k8s-integration (push) Has been cancelled
PR Test (SMG) / finish (push) Has been cancelled
PR Test (SMG) / summarize-benchmarks (push) Has been cancelled
Release SGLang Model Gateway Docker Image / publish (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on macos (aarch64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (aarch64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (x86_64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (aarch64 - musllinux_1_1) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (x86_64 - musllinux_1_1) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / Build SDist (push) Has been cancelled
Release SGLang Model Gateway to PyPI / Upload to PyPI (push) Has been cancelled
Release SGLang Kernels / build-cu129-matrix (aarch64, 12.9, 3.10, arm-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / build-cu129-matrix (x86_64, 12.9, 3.10, x64-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / release-cu129 (push) Has been cancelled
Release SGLang Kernels / build-cu130-matrix (aarch64, 13.0, 3.10, arm-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / build-cu130-matrix (x86_64, 13.0, 3.10, x64-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / release-cu130 (push) Has been cancelled
Release SGLang Kernels / build-rocm-matrix (3.10, 700) (push) Has been cancelled
Release SGLang Kernels / build-rocm-matrix (3.10, 720) (push) Has been cancelled
Release SGLang Kernels / release-rocm700 (push) Has been cancelled
Release SGLang Kernels / release-rocm720 (push) Has been cancelled
Release SGLang Kernels / build-musa43 (43, 3.10) (push) Has been cancelled
Release SGLang Kernels / release-musa43 (push) Has been cancelled
chore: import upstream snapshot with attribution
2026-07-13 12:38:16 +08:00

187 lines
7.3 KiB
Python

"""
Unit tests for weight validation and cache cleanup logic.
Tests the fix for issue #14754 - ensuring that missing shards do not trigger
entire cache deletion, which can cause race conditions in multi-process scenarios.
"""
import json
import os
import struct
import tempfile
import unittest
from sglang.srt.model_loader.ci_weight_validation import (
_check_index_files_exist,
_validate_sharded_model,
)
class TestWeightValidation(unittest.TestCase):
"""Tests for weight validation functions."""
def test_validate_sharded_model_missing_shard(self):
"""
Test that missing shards are detected correctly.
This is the core test for issue #14754 fix: when a shard is missing,
the validation should return is_valid=False with an error message
containing "Missing", but corrupted_files should be empty (indicating
this is a missing shard issue, not a corruption issue).
This distinction is critical because:
- Missing shards: should NOT delete cache (other processes may be using it)
- Corrupted files: should delete only the corrupted files selectively
"""
with tempfile.TemporaryDirectory() as tmpdir:
# Create partial shards (missing shard 3)
for i in [1, 2]: # Missing shard 3
open(
os.path.join(tmpdir, f"model-0000{i}-of-00003.safetensors"), "w"
).close()
# Create index file
index_data = {
"weight_map": {
"layer1": "model-00001-of-00003.safetensors",
"layer2": "model-00002-of-00003.safetensors",
"layer3": "model-00003-of-00003.safetensors",
}
}
with open(os.path.join(tmpdir, "model.safetensors.index.json"), "w") as f:
json.dump(index_data, f)
weight_files = [
os.path.join(tmpdir, f"model-0000{i}-of-00003.safetensors")
for i in [1, 2]
]
is_valid, error_msg, corrupted_files = _validate_sharded_model(
tmpdir, weight_files
)
self.assertFalse(is_valid)
self.assertIn("Missing", error_msg)
# CRITICAL: corrupted_files should be empty for missing shards
# This is what prevents entire cache deletion
self.assertEqual(corrupted_files, [])
def test_validate_sharded_model_all_present(self):
"""Test that complete shards pass validation."""
with tempfile.TemporaryDirectory() as tmpdir:
# Create all shards with valid safetensors header
for i in [1, 2, 3]:
filepath = os.path.join(tmpdir, f"model-0000{i}-of-00003.safetensors")
# Create a minimal valid safetensors file
# Header: 8 bytes for header size + JSON header
header = b'{"__metadata__":{}}'
header_size = len(header)
with open(filepath, "wb") as f:
f.write(struct.pack("<Q", header_size))
f.write(header)
# Create index file
index_data = {
"weight_map": {
"layer1": "model-00001-of-00003.safetensors",
"layer2": "model-00002-of-00003.safetensors",
"layer3": "model-00003-of-00003.safetensors",
}
}
with open(os.path.join(tmpdir, "model.safetensors.index.json"), "w") as f:
json.dump(index_data, f)
weight_files = [
os.path.join(tmpdir, f"model-0000{i}-of-00003.safetensors")
for i in [1, 2, 3]
]
is_valid, error_msg, corrupted_files = _validate_sharded_model(
tmpdir, weight_files
)
self.assertTrue(is_valid)
self.assertIsNone(error_msg)
self.assertEqual(corrupted_files, [])
def test_validate_sharded_model_corrupted_shard(self):
"""
Test that corrupted shards are detected and returned in corrupted_files.
This tests the other branch: when a file exists but is corrupted
(invalid safetensors format), it should be added to corrupted_files
so that selective cleanup can remove just that file.
"""
with tempfile.TemporaryDirectory() as tmpdir:
# Create shard 1 as valid
filepath1 = os.path.join(tmpdir, "model-00001-of-00003.safetensors")
header = b'{"__metadata__":{}}'
with open(filepath1, "wb") as f:
f.write(struct.pack("<Q", len(header)))
f.write(header)
# Create shard 2 as corrupted (invalid header)
filepath2 = os.path.join(tmpdir, "model-00002-of-00003.safetensors")
with open(filepath2, "wb") as f:
f.write(b"invalid data that is not a valid safetensors file")
# Create shard 3 as valid
filepath3 = os.path.join(tmpdir, "model-00003-of-00003.safetensors")
with open(filepath3, "wb") as f:
f.write(struct.pack("<Q", len(header)))
f.write(header)
# Create index file
index_data = {
"weight_map": {
"layer1": "model-00001-of-00003.safetensors",
"layer2": "model-00002-of-00003.safetensors",
"layer3": "model-00003-of-00003.safetensors",
}
}
with open(os.path.join(tmpdir, "model.safetensors.index.json"), "w") as f:
json.dump(index_data, f)
weight_files = [filepath1, filepath2, filepath3]
is_valid, error_msg, corrupted_files = _validate_sharded_model(
tmpdir, weight_files
)
self.assertFalse(is_valid)
self.assertIn("Corrupt", error_msg)
# The corrupted file should be identified
self.assertEqual(len(corrupted_files), 1)
self.assertIn("model-00002-of-00003.safetensors", corrupted_files[0])
def test_broken_index_symlink_detected(self):
"""
Test that broken index symlinks are detected and cause validation to fail.
When an index file is a symlink pointing to a non-existent blob,
validation should fail (to trigger re-download) rather than silently
continuing and causing timeout during actual loading.
"""
with tempfile.TemporaryDirectory() as tmpdir:
# Create a broken symlink for the index file
index_path = os.path.join(tmpdir, "model.safetensors.index.json")
non_existent_blob = os.path.join(tmpdir, "blobs", "nonexistent_hash")
os.symlink(non_existent_blob, index_path)
# Verify it's a broken symlink
self.assertTrue(os.path.islink(index_path))
self.assertFalse(os.path.exists(index_path))
# Check should fail for broken symlink
is_valid, error_msg = _check_index_files_exist(tmpdir)
self.assertFalse(is_valid)
self.assertIn("Broken", error_msg)
# The broken symlink should have been cleaned up
self.assertFalse(os.path.exists(index_path))
self.assertFalse(os.path.islink(index_path))
if __name__ == "__main__":
unittest.main()