[cloudstack] branch main updated: Externalize KVM Agent storage's reboot configuration (#4586)

[nvazquez]

This is an automated email from the ASF dual-hosted git repository.

nvazquez pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/cloudstack.git


The following commit(s) were added to refs/heads/main by this push:
     new 83c0b61  Externalize KVM Agent storage's reboot configuration (#4586)
83c0b61 is described below

commit 83c0b61ab20236592181ca9f6786274c0d335c90
Author: Daniel Augusto Veronezi Salvador 
<38945620+gutoveron...@users.noreply.github.com>
AuthorDate: Tue Aug 24 01:06:00 2021 -0300

    Externalize KVM Agent storage's reboot configuration (#4586)
    
    Co-authored-by: GutoVeronezi <dan...@scclouds.com.br>
---
 agent/conf/agent.properties                                       | 3 +++
 .../src/main/java/com/cloud/agent/properties/AgentProperties.java | 8 ++++++++
 .../main/java/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 2586327..784abf7 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 d04e98f..9e5f167 100644
--- a/agent/src/main/java/com/cloud/agent/properties/AgentProperties.java
+++ b/agent/src/main/java/com/cloud/agent/properties/AgentProperties.java
@@ -33,6 +33,14 @@ 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>
      * Default value: true.
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 a939abe..022b48c 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();