.NET: chore: support retries on Cosmos storage creation

.github/workflows/dotnet-cosmosdb-integration-tests.yml

@@ -0,0 +1,158 @@
+#
+# This workflow runs Cosmos DB integration tests using the Cosmos DB emulator.
+#
+
+name: dotnet-cosmosdb-integration-tests
+
+on:
+ workflow_dispatch:
+ pull_request:
+ branches: ["main", "feature*"]
+ paths:
+ - dotnet/tests/CosmosDB.IntegrationTests/**
+ - dotnet/src/Microsoft.Extensions.AI.Agents.Runtime.Storage.CosmosDB/**
+ - '.github/workflows/dotnet-cosmosdb-integration-tests.yml'
+ merge_group:
+ branches: ["main"]
+ push:
+ branches: ["main", "feature*"]
+ paths:
+ - dotnet/tests/CosmosDB.IntegrationTests/**
+ - dotnet/src/Microsoft.Extensions.AI.Agents.Runtime.Storage.CosmosDB/**
+ - '.github/workflows/dotnet-cosmosdb-integration-tests.yml'
+ schedule:
+ - cron: "0 2 * * *" # Run at 2 AM UTC daily
+
+env:
+ COSMOSDB_TESTS_USE_EMULATOR_CICD: "true"
+
+jobs:
+ build-and-test:
+ runs-on: ${{ matrix.os }}
+
+ strategy:
+ fail-fast: false
+ matrix:
+ include:
+ - { targetFramework: "net9.0", os: "ubuntu-latest", configuration: Release }
+ # - { targetFramework: "net9.0", os: "ubuntu-latest", configuration: Debug }
+
+ services:
+ cosmosdb:
+ image: mcr.microsoft.com/cosmosdb/linux/azure-cosmos-emulator:latest
+ ports:
+ - 8081:8081
+ env:
+ AZURE_COSMOS_EMULATOR_ENABLE_DATA_PERSISTENCE: "false"
+ AZURE_COSMOS_EMULATOR_PARTITION_COUNT: "20" # the more the better for stable tests
+
+ steps:
+ - uses: actions/checkout@v4
+ with:
+ persist-credentials: false
+ sparse-checkout: |
+ .
+ .github
+ dotnet
+ - name: Setup dotnet
+ uses: actions/setup-dotnet@v4.3.1
+ with:
+ global-json-file: ${{ github.workspace }}/dotnet/global.json
+ - name: Build dotnet solutions
+ shell: bash
+ run: |
+ export SOLUTIONS=$(find ./dotnet/ -type f -name "*.slnx" | tr '\n' ' ')
+ for solution in $SOLUTIONS; do
+ dotnet build $solution -c ${{ matrix.configuration }} --warnaserror
+ done
+ - name: Package install check
+ shell: bash
+ # All frameworks are only built for the release configuration, so we only run this step for the release configuration
+ # and dotnet new doesn't support net472
+ if: matrix.configuration == 'Release' && matrix.targetFramework != 'net472'
+ run: |
+ TEMP_DIR=$(mktemp -d)
+
+ export SOLUTIONS=$(find ./dotnet/ -type f -name "*.slnx" | tr '\n' ' ')
+ for solution in $SOLUTIONS; do
+ dotnet pack $solution /property:TargetFrameworks=${{ matrix.targetFramework }} -c ${{ matrix.configuration }} --no-build --no-restore --output "$TEMP_DIR/artifacts"
+ done
+
+ pushd "$TEMP_DIR"
+
+ # Create a new console app to test the package installation
+ dotnet new console -f ${{ matrix.targetFramework }} --name packcheck --output consoleapp
+
+ # Create minimal nuget.config and use only dotnet nuget commands
+ echo '<?xml version="1.0" encoding="utf-8"?><configuration><packageSources><clear /></packageSources></configuration>' > consoleapp/nuget.config
+
+ # Add sources with local first using dotnet nuget commands
+ dotnet nuget add source ../artifacts --name local --configfile consoleapp/nuget.config
+ dotnet nuget add source https://api.nuget.org/v3/index.json --name nuget.org --configfile consoleapp/nuget.config
+
+ # Change to project directory to ensure local nuget.config is used
+ pushd consoleapp
+ dotnet add packcheck.csproj package Microsoft.Extensions.AI.Agents --prerelease
+ dotnet build -f ${{ matrix.targetFramework }} -c ${{ matrix.configuration }} packcheck.csproj
+
+ # Clean up
+ popd
+ popd
+ rm -rf "$TEMP_DIR"
+
+ - name: Wait for Cosmos DB Emulator to be ready
+ run: |
+ set -e
+ for i in $(seq 1 120); do
+ if curl -sk https://localhost:8081/_explorer/emulator.pem -o /dev/null; then
+ echo "Emulator is up."
+ break
+ fi
+ echo "Waiting for emulator... ($i/120)"
+ sleep 2
+ done
+
+ - name: Install emulator TLS certificate into system trust store
+ run: |
+ set -e
+ sudo apt-get update
+ sudo apt-get install -y ca-certificates curl openssl
+ # Fetch the PEM directly from the emulator's explorer endpoint
+ curl -sk https://localhost:8081/_explorer/emulator.pem -o cosmos-emulator.crt
+ # Install with the correct .crt extension so update-ca-certificates picks it up
+ sudo cp cosmos-emulator.crt /usr/local/share/ca-certificates/cosmos-emulator.crt
+ sudo update-ca-certificates
+
+ - name: Verify TLS now trusts the emulator
+ run: |
+ # Use -servername to avoid SNI warning and check verification
+ echo | openssl s_client -connect localhost:8081 -servername localhost 2>/dev/null | grep -E "Verify return code|subject=|issuer="
+ # Expect: "Verify return code: 0 (ok)"
+
+ - name: Run Cosmos DB Integration Tests
+ shell: bash
+ run: |
+ # Run the specific CosmosDB integration tests
+ dotnet test ./dotnet/tests/CosmosDB.IntegrationTests/Microsoft.Extensions.AI.Agents.Runtime.Storage.CosmosDB.Tests/Microsoft.Extensions.AI.Agents.Runtime.Storage.CosmosDB.Tests.csproj \
+ -f ${{ matrix.targetFramework }} \
+ -c ${{ matrix.configuration }} \
+ --no-build \
+ -v Normal \
+ --logger trx \
+ --collect:"XPlat Code Coverage" \
+ --results-directory:"TestResults/Coverage/" \
+ -- DataCollectionRunSettings.DataCollectors.DataCollector.Configuration.ExcludeByAttribute=GeneratedCodeAttribute,CompilerGeneratedAttribute,ExcludeFromCodeCoverageAttribute
+
+ # Generate test reports and check coverage
+ - name: Generate test reports
+ uses: danielpalme/ReportGenerator-GitHub-Action@5.4.11
+ with:
+ reports: "./TestResults/Coverage/**/coverage.cobertura.xml"
+ targetdir: "./TestResults/Reports"
+ reporttypes: "HtmlInline;JsonSummary"
+
+ - name: Upload coverage report artifact
+ uses: actions/upload-artifact@v4
+ with:
+ name: CosmosDB-CoverageReport-${{ matrix.os }}-${{ matrix.targetFramework }}-${{ matrix.configuration }}
+ path: ./TestResults/Reports

dotnet/Directory.Packages.props

@@ -5,14 +5,18 @@
  <ManagePackageVersionsCentrally>true</ManagePackageVersionsCentrally>
  <CentralPackageTransitivePinningEnabled>true</CentralPackageTransitivePinningEnabled>
  </PropertyGroup>
+ <PropertyGroup>
+ <!-- Aspire -->
+ <AspireAppHostSdkVersion>9.4.1</AspireAppHostSdkVersion>
+ </PropertyGroup>
  <ItemGroup>
  <!-- Azure.* -->
- <PackageVersion Include="Aspire.Azure.AI.OpenAI" Version="9.3.1-preview.1.25305.6" />
- <PackageVersion Include="Aspire.Hosting.AppHost" Version="9.4.0" />
- <PackageVersion Include="Aspire.Hosting.Azure.CognitiveServices" Version="9.4.0" />
- <PackageVersion Include="Aspire.Hosting.Azure.CosmosDB" Version="9.4.0" />
- <PackageVersion Include="Aspire.Microsoft.Azure.Cosmos" Version="9.3.1" />
- <PackageVersion Include="Aspire.Hosting.Testing" Version="9.3.1" />
+ <PackageVersion Include="Aspire.Azure.AI.OpenAI" Version="9.4.1-preview.1.25408.4" />
+ <PackageVersion Include="Aspire.Hosting.AppHost" Version="9.4.1" />
+ <PackageVersion Include="Aspire.Hosting.Azure.CognitiveServices" Version="9.4.1" />
+ <PackageVersion Include="Aspire.Hosting.Azure.CosmosDB" Version="9.4.1" />
+ <PackageVersion Include="Aspire.Microsoft.Azure.Cosmos" Version="9.4.1" />
+ <PackageVersion Include="Aspire.Hosting.Testing" Version="9.4.1" />
  <PackageVersion Include="Azure.AI.Agents.Persistent" Version="1.2.0-beta.1" />
  <PackageVersion Include="Azure.AI.OpenAI" Version="2.2.0-beta.5" />
  <PackageVersion Include="Azure.Identity" Version="1.14.2" />
@@ -61,6 +65,7 @@
  <PackageVersion Include="Microsoft.Extensions.Logging.Abstractions" Version="9.0.8" />
  <PackageVersion Include="Microsoft.Extensions.Logging.Console" Version="9.0.8" />
  <PackageVersion Include="Microsoft.Extensions.Logging.Testing" Version="9.0.8" />
+ <PackageVersion Include="Microsoft.Extensions.Options" Version="9.0.8" />
  <!-- Vector Stores -->
  <PackageVersion Include="Microsoft.SemanticKernel.Connectors.InMemory" Version="1.61.0-preview" />
  <!-- Agent SDKs -->

dotnet/agent-framework-dotnet.slnx

@@ -28,6 +28,7 @@
  <Folder Name="/Solution Items/.github/workflows/">
  <File Path="../.github/workflows/dotnet-build-and-test.yml" />
  <File Path="../.github/workflows/dotnet-check-coverage.ps1" />
+ <File Path="../.github/workflows/dotnet-cosmosdb-integration-tests.yml" />
  <File Path="../.github/workflows/dotnet-format.yml" />
  </Folder>
  <Folder Name="/Solution Items/docs/" />
@@ -121,6 +122,10 @@
  <Project Path="src/Microsoft.Extensions.AI.Agents/Microsoft.Extensions.AI.Agents.csproj" />
  </Folder>
  <Folder Name="/Tests/" />
+ <Folder Name="/Tests/Cosmos/">
+ <Project Path="tests/CosmosDB.IntegrationTests/CosmosDB.Testing.AppHost/CosmosDB.Testing.AppHost.csproj" />
+ <Project Path="tests/CosmosDB.IntegrationTests/Microsoft.Extensions.AI.Agents.Runtime.Storage.CosmosDB.Tests/Microsoft.Extensions.AI.Agents.Runtime.Storage.CosmosDB.Tests.csproj" />
+ </Folder>
  <Folder Name="/Tests/IntegrationTests/">
  <Project Path="tests/AgentConformance.IntegrationTests/AgentConformance.IntegrationTests.csproj" />
  <Project Path="tests/AzureAIAgentsPersistent.IntegrationTests/AzureAIAgentsPersistent.IntegrationTests.csproj" />

dotnet/samples/AgentWebChat/AgentWebChat.AppHost/AgentWebChat.AppHost.csproj

@@ -1,6 +1,6 @@
 <Project Sdk="Microsoft.NET.Sdk">
 
- <Sdk Name="Aspire.AppHost.Sdk" Version="9.3.1" />
+ <Sdk Name="Aspire.AppHost.Sdk" Version="$(AspireAppHostSdkVersion)" />
 
  <PropertyGroup>
  <OutputType>Exe</OutputType>

dotnet/src/Microsoft.Extensions.AI.Agents.Runtime.Storage.CosmosDB/ActorDocuments.cs

@@ -11,13 +11,6 @@ namespace Microsoft.Extensions.AI.Agents.Runtime.Storage.CosmosDB;
 /// the entire actor's state for optimistic concurrency control.
 /// This document contains no actor state data. It only serves to track last modified
 /// time and provide a single ETag for the actor's state.
-///
-/// Example structure:
-/// {
-/// "id": "rootdoc", // Root document ID (constant per actor partition)
-/// "actorId": "actor-123", // Partition key (actor ID)
-/// "lastModified": "2024-...", // Timestamp
-/// }
 /// </summary>
 public sealed class ActorRootDocument
 {
@@ -27,9 +20,13 @@ public sealed class ActorRootDocument
  public string Id { get; set; } = default!;
 
  /// <summary>
- /// The actor ID.
+ /// The actor type.
  /// </summary>
- public string ActorId { get; set; } = default!;
+ public string ActorType { get; set; } = default!;
+ /// <summary>
+ /// The actor key.
+ /// </summary>
+ public string ActorKey { get; set; } = default!;
 
  /// <summary>
  /// The last modified timestamp.
@@ -55,9 +52,13 @@ public sealed class ActorStateDocument
  public string Id { get; set; } = default!;
 
  /// <summary>
- /// The actor ID.
+ /// The actor type.
+ /// </summary>
+ public string ActorType { get; set; } = default!;
+ /// <summary>
+ /// The actor key.
  /// </summary>
- public string ActorId { get; set; } = default!;
+ public string ActorKey { get; set; } = default!;
 
  /// <summary>
  /// The logical key for the state entry.

dotnet/src/Microsoft.Extensions.AI.Agents.Runtime.Storage.CosmosDB/CosmosActorStateStorage.cs

@@ -11,7 +11,7 @@ namespace Microsoft.Extensions.AI.Agents.Runtime.Storage.CosmosDB;
 /// <summary>
 /// Cosmos DB implementation of actor state storage.
 /// </summary>
-public class CosmosActorStateStorage : IActorStateStorage
+public class CosmosActorStateStorage : IActorStateStorage, IAsyncDisposable
 {
  private readonly LazyCosmosContainer _lazyContainer;
  private const string InitialEtag = "0"; // Initial ETag value when no state exists
@@ -47,8 +47,8 @@ public async ValueTask<WriteResponse> WriteStateAsync(
  }
 
  var container = await this._lazyContainer.GetContainerAsync().ConfigureAwait(false);
- var batch = container.CreateTransactionalBatch(GetPartitionKey(actorId));
- var actorIdStr = actorId.ToString();
+ var (partitionKey, actorType, actorKey) = BuildPartitionKey(actorId);
+ var batch = container.CreateTransactionalBatch(partitionKey);
 
  // Add data operations to batch
  foreach (var op in operations)
@@ -61,7 +61,8 @@ public async ValueTask<WriteResponse> WriteStateAsync(
  var item = new ActorStateDocument
  {
  Id = docId,
- ActorId = actorIdStr,
+ ActorType = actorType,
+ ActorKey = actorKey,
  Key = set.Key,
  Value = set.Value
  };
@@ -83,7 +84,8 @@ public async ValueTask<WriteResponse> WriteStateAsync(
  var newRoot = new ActorRootDocument
  {
  Id = RootDocumentId,
- ActorId = actorId.ToString(),
+ ActorType = actorType,
+ ActorKey = actorKey,
  LastModified = DateTimeOffset.UtcNow,
  };
 
@@ -103,6 +105,7 @@ public async ValueTask<WriteResponse> WriteStateAsync(
  var result = await batch.ExecuteAsync(cancellationToken).ConfigureAwait(false);
  if (!result.IsSuccessStatusCode)
  {
+ _ = result.ErrorMessage;
  return new WriteResponse(eTag: string.Empty, success: false);
  }
 
@@ -135,6 +138,8 @@ public async ValueTask<ReadResponse> ReadStateAsync(
 
  // Read root document first to get actor-level ETag
  string actorETag = await this.GetActorETagAsync(container, actorId, cancellationToken).ConfigureAwait(false);
+ var actorType = actorId.Type.ToString();
+ var actorKey = actorId.Key;
 
  foreach (var op in operations)
  {
@@ -162,14 +167,16 @@ public async ValueTask<ReadResponse> ReadStateAsync(
  QueryDefinition query;
  if (!string.IsNullOrEmpty(list.KeyPrefix))
  {
- query = new QueryDefinition("SELECT c.key FROM c WHERE c.actorId = @actorId AND c.key != null AND STARTSWITH(c.key, @keyPrefix)")
- .WithParameter("@actorId", actorId.ToString())
+ query = new QueryDefinition("SELECT c.key FROM c WHERE c.actorType = @actorType AND c.actorKey = @actorKey AND c.key != null AND STARTSWITH(c.key, @keyPrefix)")
+ .WithParameter("@actorType", actorType)
+ .WithParameter("@actorKey", actorKey)
  .WithParameter("@keyPrefix", list.KeyPrefix);
  }
  else
  {
- query = new QueryDefinition("SELECT c.key FROM c WHERE c.actorId = @actorId AND c.key != null")
- .WithParameter("@actorId", actorId.ToString());
+ query = new QueryDefinition("SELECT c.key FROM c WHERE c.actorType = @actorType AND c.actorKey = @actorKey AND c.key != null")
+ .WithParameter("@actorType", actorType)
+ .WithParameter("@actorKey", actorKey);
  }
 
  var requestOptions = new QueryRequestOptions
@@ -212,7 +219,18 @@ public async ValueTask<ReadResponse> ReadStateAsync(
  private const string RootDocumentId = "rootdoc";
 
  private static PartitionKey GetPartitionKey(ActorId actorId)
- => new(actorId.ToString());
+ {
+ var (partitionKey, _, _) = BuildPartitionKey(actorId);
+ return partitionKey;
+ }
+
+ private static (PartitionKey partitionKey, string actorType, string actorKey) BuildPartitionKey(ActorId actorId)
+ {
+ var actorType = actorId.Type.ToString();
+ var actorKey = actorId.Key;
+ var partitionKey = new PartitionKeyBuilder().Add(actorType).Add(actorKey).Build();
+ return (partitionKey, actorType, actorKey);
+ }
 
  /// <summary>
  /// Gets the current ETag for the actor's root document.
@@ -234,4 +252,13 @@ private async ValueTask<string> GetActorETagAsync(Container container, ActorId a
  return InitialEtag;
  }
  }
+
+ /// <summary>
+ /// Disposes the Cosmos DB container asynchronously.
+ /// </summary>
+ public async ValueTask DisposeAsync()
+ {
+ await this._lazyContainer.DisposeAsync().ConfigureAwait(false);
+ GC.SuppressFinalize(this);
+ }
 }