From 83c0b61ab20236592181ca9f6786274c0d335c90 Mon Sep 17 00:00:00 2001
From: Daniel Augusto Veronezi Salvador
 <38945620+GutoVeronezi@users.noreply.github.com>
Date: Tue, 24 Aug 2021 01:06:00 -0300
Subject: [PATCH] Externalize KVM Agent storage's reboot configuration (#4586)

Co-authored-by: GutoVeronezi <daniel@scclouds.com.br>
---
 agent/conf/agent.properties                               | 3 +++
 .../java/com/cloud/agent/properties/AgentProperties.java  | 8 ++++++++
 .../com/cloud/hypervisor/kvm/resource/KVMHAMonitor.java   | 4 +++-
 3 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/agent/conf/agent.properties b/agent/conf/agent.properties
index 25863276c09..784abf799d1 100644
--- a/agent/conf/agent.properties
+++ b/agent/conf/agent.properties
@@ -271,5 +271,8 @@ iscsi.session.cleanup.enabled=false
 # Depending on the use case, this timeout might need increasing/decreasing.
 # heartbeat.update.timeout=60000
 
+# This parameter specifies if the host must be rebooted when something goes wrong with the heartbeat.
+# reboot.host.and.alert.management.on.heartbeat.timeout=true
+
 # Enable manually setting CPU's topology on KVM's VM.
 # enable.manually.setting.cpu.topology.on.kvm.vm=true
diff --git a/agent/src/main/java/com/cloud/agent/properties/AgentProperties.java b/agent/src/main/java/com/cloud/agent/properties/AgentProperties.java
index d04e98fb704..9e5f167cc4e 100644
--- a/agent/src/main/java/com/cloud/agent/properties/AgentProperties.java
+++ b/agent/src/main/java/com/cloud/agent/properties/AgentProperties.java
@@ -32,6 +32,14 @@ public class AgentProperties{
     */
     public static final Property<Integer> HEARTBEAT_UPDATE_TIMEOUT = new Property<Integer>("heartbeat.update.timeout", 60000);
 
+    /**
+     * Reboot host and alert management on heartbeat timeout. <br>
+     * Data type: boolean.<br>
+     * Default value: true.
+     */
+    public static final Property<Boolean> REBOOT_HOST_AND_ALERT_MANAGEMENT_ON_HEARTBEAT_TIMEOUT
+        = new Property<Boolean>("reboot.host.and.alert.management.on.heartbeat.timeout", true);
+
     /**
      * Enable manually setting CPU's topology on KVM's VM. <br>
      * Data type: boolean.<br>
diff --git a/plugins/hypervisors/kvm/src/main/java/com/cloud/hypervisor/kvm/resource/KVMHAMonitor.java b/plugins/hypervisors/kvm/src/main/java/com/cloud/hypervisor/kvm/resource/KVMHAMonitor.java
index a939abe3bbe..022b48c929f 100644
--- a/plugins/hypervisors/kvm/src/main/java/com/cloud/hypervisor/kvm/resource/KVMHAMonitor.java
+++ b/plugins/hypervisors/kvm/src/main/java/com/cloud/hypervisor/kvm/resource/KVMHAMonitor.java
@@ -36,6 +36,7 @@ public class KVMHAMonitor extends KVMHABase implements Runnable {
 
     private static final Logger s_logger = Logger.getLogger(KVMHAMonitor.class);
     private final Map<String, NfsStoragePool> storagePool = new ConcurrentHashMap<>();
+    private final boolean rebootHostAndAlertManagementOnHeartbeatTimeout;
 
     private final String hostPrivateIp;
 
@@ -47,6 +48,7 @@ public class KVMHAMonitor extends KVMHABase implements Runnable {
         configureHeartBeatPath(scriptPath);
 
         _heartBeatUpdateTimeout = AgentPropertiesFileHandler.getPropertyValue(AgentProperties.HEARTBEAT_UPDATE_TIMEOUT);
+        rebootHostAndAlertManagementOnHeartbeatTimeout = AgentPropertiesFileHandler.getPropertyValue(AgentProperties.REBOOT_HOST_AND_ALERT_MANAGEMENT_ON_HEARTBEAT_TIMEOUT);
     }
 
     private static synchronized void configureHeartBeatPath(String scriptPath) {
@@ -134,7 +136,7 @@ public class KVMHAMonitor extends KVMHABase implements Runnable {
 
                 }
 
-                if (result != null) {
+                if (result != null && rebootHostAndAlertManagementOnHeartbeatTimeout) {
                     s_logger.warn(String.format("Write heartbeat for pool [%s] failed: %s; stopping cloudstack-agent.", uuid, result));
                     Script cmd = createHeartBeatCommand(primaryStoragePool, null, false);
                     result = cmd.execute();