ARTEMIS-4545 Allow node ID to be configured

AntonRoskvist · AntonRoskvist · commit d0e192a6d9b0 · 2024-06-03T09:44:17.000+02:00
diff --git a/artemis-server/src/main/java/org/apache/activemq/artemis/core/config/Configuration.java b/artemis-server/src/main/java/org/apache/activemq/artemis/core/config/Configuration.java
@@ -70,6 +70,16 @@ public interface Configuration {
     */
    Configuration setName(String name);
 
+   /**
+    * Returns the ID of the node.
+    */
+   String getNodeID();
+
+   /**
+    * Sets the ID of the node. If not set, a UUID generated on first startup will be used instead.
+    */
+   Configuration setNodeID(String nodeID);
+
    /**
     * We use Bean-utils to pass in System.properties that start with {@link #setSystemPropertyPrefix(String)}.
     * The default should be 'brokerconfig.' (Including the ".").
diff --git a/artemis-server/src/main/java/org/apache/activemq/artemis/core/config/impl/ConfigurationImpl.java b/artemis-server/src/main/java/org/apache/activemq/artemis/core/config/impl/ConfigurationImpl.java
@@ -153,6 +153,8 @@ public class ConfigurationImpl implements Configuration, Serializable {
 
    private String name = "localhost";
 
+   private String nodeID = null;
+
    private boolean persistenceEnabled = ActiveMQDefaultConfiguration.isDefaultPersistenceEnabled();
 
    private int maxRedeliveryRecords = ActiveMQDefaultConfiguration.getDefaultMaxRedeliveryRecords();
@@ -2559,6 +2561,17 @@ public ConfigurationImpl setName(String name) {
       return this;
    }
 
+   @Override
+   public String getNodeID() {
+      return nodeID;
+   }
+
+   @Override
+   public ConfigurationImpl setNodeID(String nodeID) {
+      this.nodeID = nodeID;
+      return this;
+   }
+
    @Override
    public ConfigurationImpl setResolveProtocols(boolean resolveProtocols) {
       this.resolveProtocols = resolveProtocols;
@@ -2707,6 +2720,7 @@ public int hashCode() {
       result = prime * result + (int) (messageExpiryScanPeriod ^ (messageExpiryScanPeriod >>> 32));
       result = prime * result + messageExpiryThreadPriority;
       result = prime * result + ((name == null) ? 0 : name.hashCode());
+      result = prime * result + ((nodeID == null) ? 0 : nodeID.hashCode());
       result = prime * result + ((outgoingInterceptorClassNames == null) ? 0 : outgoingInterceptorClassNames.hashCode());
       result = prime * result + ((pagingDirectory == null) ? 0 : pagingDirectory.hashCode());
       result = prime * result + (persistDeliveryCountBeforeDelivery ? 1231 : 1237);
@@ -2910,6 +2924,11 @@ public boolean equals(Object obj) {
             return false;
       } else if (!name.equals(other.name))
          return false;
+      if (nodeID == null) {
+         if (other.nodeID != null)
+            return false;
+      } else if (!nodeID.equals(other.nodeID))
+         return false;
       if (outgoingInterceptorClassNames == null) {
          if (other.outgoingInterceptorClassNames != null)
             return false;
diff --git a/artemis-server/src/main/java/org/apache/activemq/artemis/core/deployers/impl/FileConfigurationParser.java b/artemis-server/src/main/java/org/apache/activemq/artemis/core/deployers/impl/FileConfigurationParser.java
@@ -424,6 +424,8 @@ public void parseMainConfig(final Element e, final Configuration config) throws
 
       config.setName(getString(e, "name", config.getName(), NO_CHECK));
 
+      config.setNodeID(getString(e, "node-id", config.getNodeID(), NO_CHECK));
+
       config.setSystemPropertyPrefix(getString(e, "system-property-prefix", config.getSystemPropertyPrefix(), NOT_NULL_OR_EMPTY));
 
       NodeList haPolicyNodes = e.getElementsByTagName("ha-policy");
diff --git a/artemis-server/src/main/java/org/apache/activemq/artemis/core/server/NodeManager.java b/artemis-server/src/main/java/org/apache/activemq/artemis/core/server/NodeManager.java
@@ -83,6 +83,15 @@ public SimpleString getNodeId() {
       }
    }
 
+   /**
+    * Returns the converted form of a given nodeID
+    *
+    * @param nodeID
+    */
+   public String getConvertedNodeId(String nodeID) {
+      return new UUID(UUID.TYPE_TIME_BASED, UUID.stringToBytes(nodeID)).toString();
+   }
+
    public long readNodeActivationSequence() throws NodeManagerException {
       // TODO make it abstract
       throw new UnsupportedOperationException("TODO");
@@ -117,16 +126,14 @@ public UUID getUUID() {
    }
 
    /**
-    * Sets the nodeID.
-    * <p>
-    * Only used by replicating backups.
+    * Sets the nodeID
     *
     * @param nodeID
     */
    public void setNodeID(String nodeID) {
       synchronized (nodeIDGuard) {
-         this.nodeID = new SimpleString(nodeID);
          this.uuid = new UUID(UUID.TYPE_TIME_BASED, UUID.stringToBytes(nodeID));
+         this.nodeID = new SimpleString(uuid.toString());
       }
    }
 
diff --git a/artemis-server/src/main/java/org/apache/activemq/artemis/core/server/impl/ActiveMQServerImpl.java b/artemis-server/src/main/java/org/apache/activemq/artemis/core/server/impl/ActiveMQServerImpl.java
@@ -617,9 +617,35 @@ protected NodeManager createNodeManager(final File directory, boolean replicatin
       } else {
          manager = new FileLockNodeManager(directory, replicatingBackup, configuration.getJournalLockAcquisitionTimeout(), scheduledPool);
       }
+
+      if (!replicatingBackup && configuration.getNodeID() != null) {
+         manager.setNodeID(toCompatibleNodeID(configuration.getNodeID()));
+      }
+
       return manager;
    }
 
+   private String toCompatibleNodeID(String nodeID) {
+      if (nodeID == null) {
+         return null;
+      }
+
+      final int len = nodeID.length();
+
+      if (!(len > 0)) {
+         return null;
+      }
+
+      if (len >= 16) {
+         nodeID = nodeID.substring(0, 16);
+      } else if (len % 2 != 0) {
+         // must be even for conversion to uuid, extend to next even
+         nodeID = nodeID + "+";
+      }
+
+      return nodeID.replace('-', '.');
+   }
+
    @Override
    public OperationContext newOperationContext() {
       return getStorageManager().newContext(getExecutorFactory().getExecutor());
diff --git a/artemis-server/src/main/java/org/apache/activemq/artemis/core/server/impl/FileBasedNodeManager.java b/artemis-server/src/main/java/org/apache/activemq/artemis/core/server/impl/FileBasedNodeManager.java
@@ -208,7 +208,9 @@ protected final synchronized void createNodeId() throws IOException {
             channel.write(id, 3);
             channel.force(true);
          } else if (read != 16) {
-            setUUID(UUIDGenerator.getInstance().generateUUID());
+            if (getUUID() == null) {
+               setUUID(UUIDGenerator.getInstance().generateUUID());
+            }
             id.put(getUUID().asBytes(), 0, 16);
             id.position(0);
             channel.write(id, 3);
diff --git a/artemis-server/src/main/java/org/apache/activemq/artemis/core/server/impl/ReplicationBackupActivation.java b/artemis-server/src/main/java/org/apache/activemq/artemis/core/server/impl/ReplicationBackupActivation.java
@@ -80,7 +80,7 @@ public ReplicationBackupActivation(final ActiveMQServerImpl activeMQServer,
          // patch expectedNodeID
          final String coordinationId = policy.getPrimaryPolicy().getCoordinationId();
          if (coordinationId != null) {
-            expectedNodeID = coordinationId;
+            expectedNodeID = activeMQServer.getNodeManager().getConvertedNodeId(coordinationId);
          } else {
             final SimpleString serverNodeID = activeMQServer.getNodeID();
             if (serverNodeID == null || serverNodeID.isEmpty()) {
diff --git a/artemis-server/src/main/resources/schema/artemis-configuration.xsd b/artemis-server/src/main/resources/schema/artemis-configuration.xsd
@@ -36,6 +36,15 @@
             </xsd:annotation>
          </xsd:element>
 
+         <xsd:element name="node-id" type="xsd:string" maxOccurs="1" minOccurs="0">
+            <xsd:annotation>
+               <xsd:documentation>
+                  Node ID. If set, it will be used as an identifier when running in a cluster.
+                  Has to be unique within the cluster
+               </xsd:documentation>
+            </xsd:annotation>
+         </xsd:element>
+
          <xsd:element name="system-property-prefix" type="xsd:string" maxOccurs="1" minOccurs="0">
             <xsd:annotation>
                <xsd:documentation>
diff --git a/artemis-server/src/test/resources/ConfigurationTest-full-config.xml b/artemis-server/src/test/resources/ConfigurationTest-full-config.xml
@@ -20,6 +20,7 @@
         xsi:schemaLocation="urn:activemq ../../../../activemq-server/src/main/resources/schema/artemis-server.xsd">
    <core xmlns="urn:activemq:core">
       <name>SomeNameForUseOnTheApplicationServer</name>
+      <node-id>AUniqueIDForThisBroker</node-id>
       <resolve-protocols>false</resolve-protocols>
       <persistence-enabled>false</persistence-enabled>
       <scheduled-thread-pool-max-size>12345</scheduled-thread-pool-max-size>
diff --git a/artemis-server/src/test/resources/ConfigurationTest-xinclude-config.xml b/artemis-server/src/test/resources/ConfigurationTest-xinclude-config.xml
@@ -21,6 +21,7 @@
         xsi:schemaLocation="urn:activemq ../../../../activemq-server/src/main/resources/schema/artemis-server.xsd">
    <core xmlns="urn:activemq:core">
       <name>SomeNameForUseOnTheApplicationServer</name>
+      <node-id>AUniqueIDForThisBroker</node-id>
       <resolve-protocols>false</resolve-protocols>
       <persistence-enabled>false</persistence-enabled>
       <scheduled-thread-pool-max-size>12345</scheduled-thread-pool-max-size>
diff --git a/docs/user-manual/_book.adoc b/docs/user-manual/_book.adoc
@@ -57,6 +57,7 @@ include::metrics.adoc[leveloffset=1]
 
 //== Broker-to-Broker Connectivity
 
+include::node-id.adoc[leveloffset=1]
 include::core-bridges.adoc[leveloffset=1]
 include::clusters.adoc[leveloffset=1]
 include::federation.adoc[leveloffset=1]
diff --git a/docs/user-manual/node-id.adoc b/docs/user-manual/node-id.adoc
@@ -0,0 +1,40 @@
+= Node ID
+:idprefix:
+:idseparator: -
+
+When connecting multiple Artemis brokers together they can cooperate to solve a common task
+like xref:clusters.adoc[clustering] or xref:ha.adoc[high availability] for example.
+
+Implementation of coordination and control over which broker does what will vary between the different "Broker-to-broker Connectivity"
+variants, but one important aspect is the use of a unique identifier called `node-id`.
+
+By default, the brokers `node-id` will get generated the first time it starts up. The value is then
+persisted into the `journal`, on a file called `server.lock`
+
+This id _must_ be unique among all brokers to ensure proper functionality.
+
+
+== Manually setting the brokers ID
+
+In certain cases, being able to manually set a `node-id` rather than generating one can be preferable. Brokers running in
+an environment where their storage is ephemeral would be one such case. This is because any time a broker is restarted, moved
+or upgraded it could end up with a new disk and therefore also a new journal. Starting this broker up, while being configured
+identiaclly, would still give it a new `node-id`. In some configurations, this means the broker won't be able to resume
+whatever role it orinially had because the other broker it was connected to prior will look for the original `node-id`.
+
+To handle these scenarios `node-id` can be set in the `broker.xml` configuration file.
+[,xml]
+----
+<node-id>myUniqueID</node-id>
+----
+
+The selected nodeID will get converted internally into a 16-byte UUID. Therefore it might not be recognizeable to
+someone looking for it in logs or the console.
+
+The `node-id` has to be set before first starting the broker as this value will get persisted in the brokers `journal`.
+Once persisted, this is the value that the broker will use regardless of configuration.
+
+[WARNING]
+====
+If you choose to set `node-id` manually, it's uniqueness among other brokers are _UTMOST_ important.
+====
diff --git a/tests/artemis-test-support/src/main/java/org/apache/activemq/artemis/tests/util/ActiveMQTestBase.java b/tests/artemis-test-support/src/main/java/org/apache/activemq/artemis/tests/util/ActiveMQTestBase.java
@@ -455,6 +455,7 @@ protected ConfigurationImpl createBasicConfig() throws Exception {
 
    protected ConfigurationImpl createBasicConfig(final int serverID) {
       ConfigurationImpl configuration = new ConfigurationImpl().setSecurityEnabled(false).setJournalMinFiles(2).setJournalFileSize(100 * 1024).setJournalType(getDefaultJournalType()).setJournalDirectory(getJournalDir(serverID, false)).setBindingsDirectory(getBindingsDir(serverID, false)).setPagingDirectory(getPageDir(serverID, false)).setLargeMessagesDirectory(getLargeMessagesDir(serverID, false)).setJournalCompactMinFiles(0).setJournalCompactPercentage(0).setClusterPassword(CLUSTER_PASSWORD).setJournalDatasync(false);
+      configuration.setNodeID("node-" + serverID);
 
       // When it comes to the testsuite, we don't need any batching, I will leave some minimal batching to exercise the codebase
       configuration.setJournalBufferTimeout_AIO(100).setJournalBufferTimeout_NIO(100);
diff --git a/tests/integration-tests/src/test/java/org/apache/activemq/artemis/tests/integration/replication/SharedNothingReplicationTest.java b/tests/integration-tests/src/test/java/org/apache/activemq/artemis/tests/integration/replication/SharedNothingReplicationTest.java
@@ -269,6 +269,7 @@ protected Configuration createPrimaryConfiguration() throws Exception {
       conf.addClusterConfiguration(ccconf);
 
       conf.setSecurityEnabled(false).setJMXManagementEnabled(false).setJournalType(JournalType.MAPPED).setJournalFileSize(1024 * 512).setConnectionTTLOverride(60_000L);
+      conf.setNodeID("localhost::primary");
 
       return conf;
    }
@@ -300,6 +301,7 @@ protected Configuration createBackupConfiguration() throws Exception {
       conf.addClusterConfiguration(ccconf);
 
       conf.setSecurityEnabled(false).setJMXManagementEnabled(false).setJournalType(JournalType.MAPPED).setJournalFileSize(1024 * 512).setConnectionTTLOverride(60_000L);
+      conf.setNodeID("localhost::backup");
 
       return conf;
    }

Original file line number	Diff line number	Diff line change
`@@ -83,6 +83,15 @@ public SimpleString getNodeId() {`
`83`	`83`	`}`
`84`	`84`	`}`
`85`	`85`
	`86`	`+ /**`
	`87`	`+ * Returns the converted form of a given nodeID`
	`88`	`+ *`
	`89`	`+ * @param nodeID`
	`90`	`+ */`
	`91`	`+ public String getConvertedNodeId(String nodeID) {`
	`92`	`+ return new UUID(UUID.TYPE_TIME_BASED, UUID.stringToBytes(nodeID)).toString();`
	`93`	`+ }`
	`94`	`+`
`86`	`95`	`public long readNodeActivationSequence() throws NodeManagerException {`
`87`	`96`	`// TODO make it abstract`
`88`	`97`	`throw new UnsupportedOperationException("TODO");`
`@@ -117,16 +126,14 @@ public UUID getUUID() {`
`117`	`126`	`}`
`118`	`127`
`119`	`128`	`/**`
`120`		`- * Sets the nodeID.`
`121`		`- * <p>`
`122`		`- * Only used by replicating backups.`
	`129`	`+ * Sets the nodeID`
`123`	`130`	`*`
`124`	`131`	`* @param nodeID`
`125`	`132`	`*/`
`126`	`133`	`public void setNodeID(String nodeID) {`
`127`	`134`	`synchronized (nodeIDGuard) {`
`128`		`- this.nodeID = new SimpleString(nodeID);`
`129`	`135`	`this.uuid = new UUID(UUID.TYPE_TIME_BASED, UUID.stringToBytes(nodeID));`
	`136`	`+ this.nodeID = new SimpleString(uuid.toString());`
`130`	`137`	`}`
`131`	`138`	`}`
`132`	`139`