Replace custom Raft protocol with Apache Ratis

.github/workflows/e2e-ha.yml

@@ -0,0 +1,69 @@
+name: E2E HA Tests
+
+on:
+  workflow_dispatch:
+  schedule:
+    - cron: "0 0 * * *" # Runs daily at midnight
+  pull_request:
+    branches:
+      - main
+
+jobs:
+  setup:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+      packages: write
+      attestations: write
+      id-token: write
+
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+      - name: Ensure SHA pinned actions
+        uses: zgosalvez/github-actions-ensure-sha-pinned-actions@471d5ace1f08e3c4df1c4c2f7e6341aa75da434a # v5.0.3
+      - name: Run pre-commit
+        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
+        with:
+          python-version: "3.13.0"
+          cache: "pip"
+      - uses: pre-commit/action@2c7b3805fd2a0fd8c1884dcaebf91fc102a13ecd # v3.0.1
+
+      - name: Set up JDK 21
+        uses: actions/setup-java@be666c2fcd27ec809703dec50e508c2fdc7f6654 # v5.2.0
+        with:
+          distribution: "temurin"
+          java-version: 21
+
+      - name: Cache local Maven repository
+        uses: actions/cache@668228422ae6a00e4ad889ee87cd7109ec5666a7 # v5.0.4
+        with:
+          path: ~/.m2/repository
+          key: ${{ runner.os }}-maven-${{ hashFiles('**/pom.xml') }}
+          restore-keys: |
+            ${{ runner.os }}-maven-
+
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@ce360397dd3f832beb865e1373c09c0e9f86d70a # v4.0.0
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@4d04d5d9486b7bd6fa91e7baf45bbb4f8b9deedd # v4.0.0
+
+      - name: Build and package with Maven Docker profile
+        run: ./mvnw clean install -Pdocker -DskipTests --batch-mode --errors --show-version
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Run HA Tests
+        run: ./mvnw verify -DskipTests -Pintegration --batch-mode --errors --fail-never --show-version -pl e2e-ha
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Tests Reporter
+        uses: dorny/test-reporter@a43b3a5f7366b97d083190328d2c652e1a8b6aa2 # v3.0.0
+        if: success() || failure()
+        with:
+          name: IT Tests Report
+          path: "**/failsafe-reports/TEST*.xml"
+          list-tests: "failed"
+          list-suites: "failed"
+          reporter: java-junit

ATTRIBUTIONS.md

@@ -165,6 +165,21 @@ The following table lists runtime dependencies bundled with ArcadeDB distributio
 | io.undertow | undertow-core | ~2.3.x | Apache 2.0 | https://undertow.io/ |
 | io.netty | netty-* | ~4.1.x | Apache 2.0 | https://netty.io/ |
 
+### Apache Ratis (High Availability)
+
+| Group ID | Artifact ID | Version | License | Homepage |
+|----------|-------------|---------|---------|----------|
+| org.apache.ratis | ratis-server | 3.2.2   | Apache 2.0 | https://ratis.apache.org/ |
+| org.apache.ratis | ratis-grpc | 3.2.2   | Apache 2.0 | https://ratis.apache.org/ |
+| org.apache.ratis | ratis-common | 3.2.2   | Apache 2.0 | https://ratis.apache.org/ |
+| org.apache.ratis | ratis-client | 3.2.2   | Apache 2.0 | https://ratis.apache.org/ |
+| org.apache.ratis | ratis-server-api | 3.2.2   | Apache 2.0 | https://ratis.apache.org/ |
+| org.apache.ratis | ratis-proto | 3.2.2   | Apache 2.0 | https://ratis.apache.org/ |
+| org.apache.ratis | ratis-metrics-default | 3.2.2   | Apache 2.0 | https://ratis.apache.org/ |
+| org.apache.ratis | ratis-thirdparty-misc | 1.0.10  | Apache 2.0 | https://ratis.apache.org/ |
+
+**Apache Ratis Notice:** Apache Ratis is a Java library for building fault-tolerant distributed systems using the Raft consensus algorithm. It provides the foundation for ArcadeDB's High Availability clustering. Apache Ratis, Ratis, Apache, the Apache feather, and the Apache Ratis project logo are trademarks of The Apache Software Foundation.
+
 ### Apache TinkerPop / Gremlin (Optional Module)
 
 | Group ID | Artifact ID | Version | License | Homepage |

CLAUDE.md

@@ -162,7 +162,14 @@ cd package
   - `DatabaseAbstractHandler.java` — base handler (wraps commands in transactions)
   - `PostCommandHandler.java` — POST /command endpoint
   - `PostQueryHandler.java`, `GetQueryHandler.java` — query endpoints
-- **HA**: `server/src/main/java/com/arcadedb/server/ha/`
+- **HA (Ratis)**: `ha-raft/src/main/java/com/arcadedb/server/ha/raft/`
+  - `RaftHAServer.java` - Ratis server lifecycle, peer management, cluster token, health monitor
+  - `ArcadeDBStateMachine.java` - Ratis state machine for WAL replication on followers
+  - `RaftGroupCommitter.java` - batched group commit over Raft (amortizes gRPC round-trips)
+  - `RaftLogEntryCodec.java` - binary serialization for Raft log entries
+  - `SnapshotHttpHandler.java` - HTTP endpoint for full database snapshot download
+  - `ClusterMonitor.java` - follower lag tracking and alerts
+  - `HALog.java` - verbose HA logging (`arcadedb.ha.logVerbose=0/1/2/3`)
 - **Security**: `server/src/main/java/com/arcadedb/server/security/`
 
 ### Test Locations (by module)
@@ -206,10 +213,20 @@ cd package
 - **Index System**: `com.arcadedb.index.*` - LSM-Tree indexes, full-text, vector indexes
 - **Graph Engine**: `com.arcadedb.graph.*` - Vertex/Edge management, graph traversals
 - **Serialization**: `com.arcadedb.serializer.*` - Binary serialization, JSON handling
+- **Remote Client**: `com.arcadedb.remote.*` (module: `network/`) - `RemoteDatabase` / `RemoteServer` / `RemoteSchema` wrap the HTTP API. `RemoteDatabase` is **thread-safe for sharing across threads**: `RemoteSchema` uses a synchronized `reload()` with volatile snapshot-swap, and `RemoteHttpComponent.httpCommand` is stateless per call. See `server/src/test/java/com/arcadedb/server/RemoteSchemaConcurrentInitIT.java` for the regression test.
+  - `RemoteHttpComponent` wraps `HttpClient.sendAsync` in a watchdog that enforces `timeout + NETWORK_HTTP_CLIENT_WATCHDOG_SLACK` as a hard wall-clock bound (defense against JDK HttpClient stuck HTTP/2 streams).
+  - Non-idempotent methods (POST/PUT/DELETE) are **not** auto-retried on raw `IOException` because the request may have been committed with only the response lost; retrying would duplicate the write. `NeedRetryException` (explicitly declared by the server as "I did not commit") is still retried for all methods.
 
 ### Server Components
 - **HTTP API**: `com.arcadedb.server.http.*` - REST endpoints, request handling
-- **High Availability**: `com.arcadedb.server.ha.*` - Clustering, replication, leader election
+- **High Availability**: `com.arcadedb.server.ha.raft.*` (module: `ha-raft/`) - Clustering via Apache Ratis, WAL replication, leader election
+  - `RaftHAServer.java` - Ratis server lifecycle, gRPC transport, peer management, cluster token
+  - `ArcadeDBStateMachine.java` - Ratis state machine for WAL replication on followers
+  - `RaftGroupCommitter.java` - batched group commit over Raft (amortizes gRPC round-trips)
+  - `RaftLogEntryCodec.java` - binary serialization for Raft log entries
+  - `SnapshotHttpHandler.java` - HTTP endpoint for full database snapshot download
+  - `ClusterMonitor.java` - follower lag tracking and alerts
+  - `HALog.java` - verbose HA logging (`arcadedb.ha.logVerbose=0/1/2/3`: off/basic/detailed/trace)
 - **Security**: `com.arcadedb.server.security.*` - Authentication, authorization
 - **Monitoring**: `com.arcadedb.server.monitor.*` - Metrics, query profiling, health checks
 - **MCP**: `com.arcadedb.server.mcp.*` - Model Context Protocol server support

NOTICE

@@ -81,6 +81,9 @@ Copyright 2012-2021 The ANTLR Project
 Apache TinkerPop (Gremlin)
 Copyright 2015-2024 The Apache Software Foundation
 
+Apache Ratis
+Copyright 2017-2026 The Apache Software Foundation
+
 Apache Commons Compress
 Copyright 2002-2024 The Apache Software Foundation
 

bolt/src/main/java/com/arcadedb/bolt/BoltNetworkListener.java

@@ -23,7 +23,7 @@
 import com.arcadedb.log.LogManager;
 import com.arcadedb.server.ArcadeDBServer;
 import com.arcadedb.server.ServerException;
-import com.arcadedb.server.ha.network.ServerSocketFactory;
+import com.arcadedb.server.network.ServerSocketFactory;
 
 import java.io.IOException;
 import java.io.InputStream;

bolt/src/main/java/com/arcadedb/bolt/BoltProtocolPlugin.java

@@ -22,7 +22,7 @@
 import com.arcadedb.GlobalConfiguration;
 import com.arcadedb.server.ArcadeDBServer;
 import com.arcadedb.server.ServerPlugin;
-import com.arcadedb.server.ha.network.DefaultServerSocketFactory;
+import com.arcadedb.server.network.DefaultServerSocketFactory;
 
 /**
  * Server plugin that enables Neo4j BOLT protocol support.