Use milvus docker container for integration test (#443)

Signed-off-by: yhmo <yihua.mo@zilliz.com>

Author: yhmo

.github/workflows/main.yaml

@@ -45,10 +45,6 @@ jobs:
             version: 22.04
             image: ubuntu:22.04
             key: u2204
-          - distro: Fedora
-            version: 38
-            image: fedora:38
-            key: fc38
           - distro: Fedora
             version: 39
             image: fedora:39
@@ -81,7 +77,35 @@ jobs:
       - name: Unit Testing
         run: |
           make test
-  st-ubuntu:
+  integration-test:
+    name: Integration test
+    runs-on: ubuntu-22.04
+    timeout-minutes: 75
+    env:
+      CCACHE_DIR: ${{ github.workspace }}/.ccache
+      CCACHE_COMPILERCHECK: content
+      CCACHE_COMPRESS: 1
+      CCACHE_COMPRESSLEVEL: 5
+      CCACHE_MAXSIZE: 2G
+    steps:
+      - name: Env
+        run: echo "${HOME}/.local/bin" >> $GITHUB_PATH
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Cache ccache
+        uses: actions/cache@v4
+        with:
+          path: |
+            ${{ github.workspace }}/.ccache
+          key: ubuntu-2204-st-ccache-${{ github.sha }}
+          restore-keys: ubuntu-2204-st-ccache-
+      - name: Prepare
+        run: |
+          sh scripts/install_deps.sh
+      - name: Integration Testing
+        run: |
+          make st
+  coverage-ubuntu:
     name: Test all with coverage
     runs-on: ubuntu-${{ matrix.ubuntu }}
     timeout-minutes: 120
@@ -100,13 +124,14 @@ jobs:
         run: echo "${HOME}/.local/bin" >> $GITHUB_PATH
       - name: Checkout
         uses: actions/checkout@v4
-      - name: Cache ccache
+      - name: Cache ccache and conan
         uses: actions/cache@v4
         with:
           path: |
             ${{ github.workspace }}/.ccache
-          key: ubuntu-${{ matrix.ubuntu }}-ccache-${{ github.sha }}
-          restore-keys: ubuntu-${{ matrix.ubuntu }}-ccache-
+            ~/.conan2
+          key: ubuntu-${{ matrix.ubuntu }}-cache-${{ github.sha }}
+          restore-keys: ubuntu-${{ matrix.ubuntu }}-cache-
       - uses: actions/setup-python@v4
         with:
           python-version: '3.8'
@@ -186,7 +211,7 @@ jobs:
         shell: cmd
         run: |
           call "%ProgramFiles%\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat"
-          cmake -S . -B build -DMILVUS_BUILD_TEST=YES -G Ninja
+          cmake -S . -B build -DMILVUS_BUILD_TEST=YES -G Ninja -DCMAKE_C_COMPILER_LAUNCHER=ccache -DCMAKE_CXX_COMPILER_LAUNCHER=ccache
           cmake --build build
       - name: Unit Testing
         shell: cmd

CHANGELOG.md

@@ -85,7 +85,7 @@
 
 ## milvus-sdk-cpp 2.5.2 (2025-12-04)
 ### Feature
-- Introduce MilvusClientV2 for clean/expanable interfaces
+- Introduce MilvusClientV2 for clean/expandable interfaces
 
 ## milvus-sdk-cpp 2.5.1 (2025-11-21)
 ### Feature
@@ -154,10 +154,10 @@
 - Support search with grouping by field
 - Add retry machinery for rpc call
 - Add CurrentUsedDatabase method for MilvusClient
-- Add methods for QuertResults/SearchResults to return row-based results
+- Add methods for QueryResults/SearchResults to return row-based results
 
 
-### Orther changes
+### Other changes
 - Remove unsupported index types(IVF_HNSW/RHNSW_FLAT/RHNSW_SQ/RHNSW_PQ/ANNOY)
 - Remove unsupported metric types(TANIMOTO/SUBSTRUCTURE/SUPERSTRUCTURE)
 - Remove CalcDistance interface, not supported by milvus 2.4

DEVELOPMENT.md

@@ -56,7 +56,7 @@ $ make
 
 By default, `make` command downloads gRPC source code from github and compile it under the CMake build directory.
 
-You can use a pre-built gRPC lib build by youself. milvus-sdk-cpp 2.4 is using gRPC v1.59.0, make sure your gRPC version is compatible.
+You can use a pre-built gRPC lib build by yourself. milvus-sdk-cpp 2.4 is using gRPC v1.59.0, make sure your gRPC version is compatible.
 
 ### Download gRPC source code
 ```shell
@@ -80,7 +80,7 @@ $ cmake -DCMAKE_INSTALL_PREFIX=/path/to/pre-installed/grpc  -DBUILD_SHARED_LIBS=
 $ make
 $ make install
 ```
-Make sure the `BUILD_SHARED_LIBS` is `ON` since milvus-sdk-cpp dynamiclly links to gRPC.
+Make sure the `BUILD_SHARED_LIBS` is `ON` since milvus-sdk-cpp dynamically links to gRPC.
 
 ### Use `GRPC_PATH` to specify the external gRPC and build milvus-sdk-cpp
 ```shell
@@ -113,7 +113,7 @@ to automatic format all c++ source code
 ## Run tests, and add testing code
 Milvus C++ SDK using googletest as a test framework. You could run `make test` to run unit testing and integration testing.
 
-If you have an pre-installed gRPC, use `GRPC_PATH` to specify the path:
+If you have a pre-installed gRPC, use `GRPC_PATH` to specify the path:
 ```shell
 $ make test GRPC_PATH=/path/to/pre-installed/grpc
 ```
@@ -132,8 +132,8 @@ $ ./cmake_build/test/testing-st
 ```
 
 ### Run acceptance/system tests with real Milvus server
-The acceptance/system tests are not included by default. You cloud using the below commands to run them:
-- `make st` user the top source directory
+The acceptance/system tests are not included by default. You could use the below commands to run them:
+- `make st` under the top source directory
 - `make system-test` under the CMake build directory
 
 The acceptance/system tests will startup container by docker, and using jq to capture the output from docker inspect,
@@ -154,7 +154,7 @@ After the command, a folder named "code_coverage" will be created under the proj
 You could open the **code_coverage/index.html** by a web browser to review the code coverage report.
 
 ## Generate documentation
-Milvus C++ SDK using **doxygen** tool to generate documentation. Run `make documentation` to generate documentation.
+Milvus C++ SDK uses **doxygen** tool to generate documentation. Run `make documentation` to generate documentation.
 Typically, we only publish documentation before releasing a new sdk version.
 Since the **doxygen** is not included in the `install_deps.sh`, you need to manually install it if you want to generate the documentation by yourself.
 
@@ -165,7 +165,7 @@ Since the **doxygen** is not included in the `install_deps.sh`, you need to manu
 The setup steps and development environment for macOS are similar to Linux.
 You could use `install_deps.sh` to install dependencies and use the same `make` commands for build, lint, test, and coverage.
 
-### Prerequests
+### Prerequisites
 Before you run `install_deps.sh` to install dependencies, you should make sure:
-- Already installed (Homebrew)[https://brew.sh/]
-- Install Command line tools for Xcode, by command: `xcode-select --instal`
+- Already installed [Homebrew](https://brew.sh/)
+- Install Command line tools for Xcode, by command: `xcode-select --install`

examples/example.md

@@ -47,7 +47,7 @@ Examples for MilvusClientV2:
 - `./cmake_build/examples/v2/sdk_nullable_field_v2`: example to show the usage of nullable field.
 - `./cmake_build/examples/v2/sdk_partition_key_v2`: example to show the usage of partition key.
 - `./cmake_build/examples/v2/sdk_rbac_v2`: example to show the usage of RBAC.
-- `./cmake_build/examples/v2/sdk_rerank_function_v2`: example to show the usage of rarank function.
+- `./cmake_build/examples/v2/sdk_rerank_function_v2`: example to show the usage of rerank function.
 - `./cmake_build/examples/v2/sdk_run_analyzer_v2`: example to show the usage of run_analyzer().
 - `./cmake_build/examples/v2/sdk_simple_v2`: example to quickly create a simple collection.
 - `./cmake_build/examples/v2/sdk_struct_field_v2`: example to show the usage of struct field.

scripts/build.sh

@@ -138,6 +138,10 @@ if [[ "${BUILD_FROM_CONAN}" == "ON" ]]; then
 
   # Build folder layout follows Conan 2 CMakeToolchain defaults:
   #   cmake_build/build/<BuildType>/generators/conan_toolchain.cmake
+  # Workaround: CMake 4.x removed compat with cmake_minimum_required < 3.5,
+  # which breaks older Conan recipes (e.g. c-ares/1.19.1).
+  export CMAKE_POLICY_VERSION_MINIMUM=3.5
+
   ${CONAN} install .. \
     -of . \
     -s build_type=${BUILD_TYPE} \
@@ -237,13 +241,15 @@ fi
 
 if [[ "${UNIT_TEST}" == "ON" ]]; then
   make -j ${JOBS}  || exit 1
-  ./test/testing-ut || exit 1
-  ./test/testing-it || exit 1
+  # Suppress gRPC verbose logs during tests
+  GRPC_VERBOSITY=ERROR ./test/testing-ut || exit 1
+  GRPC_VERBOSITY=ERROR ./test/testing-it || exit 1
 fi
 
 if [[ "${SYS_TEST}" == "ON" ]]; then
   make -j ${JOBS}  || exit 1
-  ./test/testing-st || exit 1
+  # Suppress gRPC verbose logs during tests
+  GRPC_VERBOSITY=ERROR ./test/testing-st || exit 1
 fi
 
 if [[ "${DO_INSTALL}" == "ON" ]]; then

scripts/install_deps.sh

@@ -96,6 +96,9 @@ install_deps_for_ubuntu_common() {
                        libssl-dev iwyu lcov git python3-pip clang-format clang-tidy
     install_linux_cmake_clang_toolchain
     install_conan
+
+    echo 'Pip install docker for integration tests.'
+    pip3 install --user "docker>=7.0.0"
 }
 
 install_deps_for_ubuntu_1804() {
@@ -115,6 +118,9 @@ install_deps_for_fedora_common() {
     ${SUDO} dnf -y install gcc gcc-c++ python2 gpg wget ccache make openssl-devel which lcov git rpm-build python3-pip perl perl-core
     install_linux_cmake_clang_toolchain
     install_conan
+
+    echo 'Pip install docker for integration tests.'
+    pip3 install --user "docker>=7.0.0"
 }
 
 install_deps_for_centos_8() {

scripts/run_clang_tidy.py

@@ -22,6 +22,7 @@
 import lintutils
 from subprocess import PIPE
 import sys
+import time
 from functools import partial
 
 
@@ -113,6 +114,10 @@ def _check_all(cmd, filenames):
         '-p',
         arguments.compile_commands
     ]
+
+    print("clang-tidy: {} files to check".format(len(linted_filenames)))
+    start_time = time.time()
+
     if arguments.fix:
         cmd.append('-fix')
         results = lintutils.run_parallel(
@@ -123,3 +128,6 @@ def _check_all(cmd, filenames):
 
     else:
         _check_all(cmd, linted_filenames)
+
+    elapsed = time.time() - start_time
+    print("clang-tidy: finished in {:.1f} seconds".format(elapsed))

src/impl/MilvusClientV2Impl.cpp

@@ -112,7 +112,7 @@ Status
 MilvusClientV2Impl::CreateCollection(const CreateCollectionRequest& request) {
     const auto& schemaPtr = request.CollectionSchema();
     if (schemaPtr == nullptr) {
-        return {StatusCode::INVALID_AGUMENT, "Collection schema is null"};
+        return {StatusCode::INVALID_ARGUMENT, "Collection schema is null"};
     }
 
     CollectionSchema& schema = *schemaPtr;
@@ -188,7 +188,7 @@ MilvusClientV2Impl::CreateCollection(const CreateSimpleCollectionRequest& reques
     if (request.PrimaryFieldType() == DataType::VARCHAR) {
         pk_field.SetMaxLength(static_cast<uint32_t>(request.MaxLength()));
     } else if (request.PrimaryFieldType() != DataType::INT64) {
-        return {StatusCode::INVALID_AGUMENT, "Primary field type is illegal"};
+        return {StatusCode::INVALID_ARGUMENT, "Primary field type is illegal"};
     }
 
     milvus::FieldSchema vector_field = milvus::FieldSchema(request.VectorFieldName(), DataType::FLOAT_VECTOR);
@@ -1062,7 +1062,7 @@ MilvusClientV2Impl::Insert(const InsertRequest& request, InsertResponse& respons
         const auto& fields = request.ColumnsData();
         const auto& rows = request.RowsData();
         if (!fields.empty() && !rows.empty()) {
-            return Status{StatusCode::INVALID_AGUMENT, "Not allow to set ColumnsData and RowsData both"};
+            return Status{StatusCode::INVALID_ARGUMENT, "Not allow to set ColumnsData and RowsData both"};
         }
 
         if (!rows.empty()) {
@@ -1161,7 +1161,7 @@ MilvusClientV2Impl::Upsert(const UpsertRequest& request, UpsertResponse& respons
         const auto& fields = request.ColumnsData();
         const auto& rows = request.RowsData();
         if (!fields.empty() && !rows.empty()) {
-            return Status{StatusCode::INVALID_AGUMENT, "Not allow to set ColumnsData and RowsData both"};
+            return Status{StatusCode::INVALID_ARGUMENT, "Not allow to set ColumnsData and RowsData both"};
         }
 
         if (!rows.empty()) {
@@ -1255,12 +1255,12 @@ MilvusClientV2Impl::Delete(const DeleteRequest& request, DeleteResponse& respons
         rpc_request.set_partition_name(request.PartitionName());
 
         if (request.Filter().empty() && request.IDs().GetRowCount() == 0) {
-            return Status{StatusCode::INVALID_AGUMENT,
+            return Status{StatusCode::INVALID_ARGUMENT,
                           "Deletion condition must be specified, by primary keys or by filter expression"};
         }
 
         if (!request.Filter().empty() && request.IDs().GetRowCount() != 0) {
-            return Status{StatusCode::INVALID_AGUMENT,
+            return Status{StatusCode::INVALID_ARGUMENT,
                           "Ambiguous filter parameter, only one deletion condition can be specified"};
         }
 

src/impl/MilvusConnection.cpp

@@ -158,7 +158,7 @@ MilvusConnection::Connect(const ConnectParam& param) {
     ss << std::put_time(local_time, "%Y-%m-%d %H:%M:%S");
     client_info->set_local_time(ss.str());
 
-    // the defalut value of ConnectTimeout is 10 seconds, means if the server could not return response
+    // the default value of ConnectTimeout is 10 seconds, means if the server could not return response
     // in 10 seconds, the MilvusClient will return an error
     ::grpc::ClientContext context;
     if (param.ConnectTimeout() > 0) {

src/impl/types/EmbeddingList.cpp

@@ -218,7 +218,7 @@ template <typename T, typename V>
 Status
 EmbeddingList::addVector(DataType data_type, const V& vector) {
     if (target_vectors_ != nullptr && target_vectors_->Type() != data_type) {
-        return {StatusCode::INVALID_AGUMENT, "Target vector must be the same type!"};
+        return {StatusCode::INVALID_ARGUMENT, "Target vector must be the same type!"};
     }
 
     StatusCode code = StatusCode::OK;
@@ -233,7 +233,7 @@ EmbeddingList::addVector(DataType data_type, const V& vector) {
         if (data_type != DataType::VARCHAR && data_type != DataType::SPARSE_FLOAT_VECTOR && dim_ != dim) {
             std::string msg =
                 "Vector size mismatch, first: " + std::to_string(dim_) + ", current: " + std::to_string(dim);
-            return {StatusCode::INVALID_AGUMENT, msg};
+            return {StatusCode::INVALID_ARGUMENT, msg};
         }
 
         std::shared_ptr<T> vectors = std::static_pointer_cast<T>(target_vectors_);
@@ -251,7 +251,7 @@ template <typename T, typename V>
 Status
 EmbeddingList::setVectors(DataType data_type, std::vector<V>&& vectors) {
     if (vectors.empty()) {
-        return {StatusCode::INVALID_AGUMENT, "Vector list is empty"};
+        return {StatusCode::INVALID_ARGUMENT, "Vector list is empty"};
     }
 
     // this method will reset the vector list
@@ -266,7 +266,7 @@ EmbeddingList::setVectors(DataType data_type, std::vector<V>&& vectors) {
         if (data_type != DataType::VARCHAR && data_type != DataType::SPARSE_FLOAT_VECTOR && dim_ != dim) {
             std::string msg =
                 "Vector size mismatch, first: " + std::to_string(dim_) + ", current: " + std::to_string(dim);
-            status = {StatusCode::INVALID_AGUMENT, msg};
+            status = {StatusCode::INVALID_ARGUMENT, msg};
         }
     }
     target_vectors_ = std::make_shared<T>("", std::move(vectors));