Providing application based monitoring and recovery for a hypervisor of an ha cluster

Abstract

An embodiment of the invention is directed to a method associated with a node comprising a hypervisor and guest VMs, each guest VM being managed by the hypervisor and disposed to run applications, the node being joined with other nodes to form an HA cluster. The method includes establishing an internal bidirectional communication channel between each guest VM and the hypervisor, and further includes sending commands and responses thereto through the channel, wherein respective commands manage a specified application running on the given guest VM. The messages are selectively monitored, to detect a failure condition associated with the specified application running on the given guest VM. Responsive to detecting a failure condition, action is taken to correct the failure condition, wherein the action includes sending at least one command through the internal channel from the hypervisor to the given guest VM.

What is claimed is: 
     
         1 . In association with a first node comprising a hypervisor and one or more guest virtual machines (VMs), wherein each guest VM is disposed to run one or more applications, and the first node can be joined with one or more other nodes to form a high availability (HA) cluster, a method comprising the steps of:
 establishing an internal bidirectional communications channel between each guest VM and the hypervisor of said first node;   sending messages that include commands and responses to commands through the internal channel, between the hypervisor and a given guest VM, wherein respective commands are sent to manage a specified application running on the given guest VM;   selectively monitoring said messages, to detect an occurrence of a failure condition associated with the specified application running on the given guest VM; and   responsive to detecting a failure condition, taking action to correct the failure condition, wherein said action includes sending at least one command through the internal channel from the hypervisor to the given guest VM.   
     
     
         2 . The method of  claim 1 , wherein:
 each of the internal bidirectional channels is established between an application availability manager associated with one of the guest VMs, and the hypervisor of said first node.   
     
     
         3 . The method of  claim 1 , wherein:
 said action to correct comprises, selectively, either sending a sequence of commands through the internal channel, from the hypervisor to the given guest VM, to stop the specified application running on the given guest VM, and to then restart the specified application on the given guest VM and in the first node or implementing failover of the specified application to a guest VM of the first node other than the given guest VM, and running the specified application on said other guest VM.   
     
     
         4 . The method of  claim 3 , wherein:
 prior to a failover to said other guest VM, a specified component of said other guest VM is operated as a replication receiver, and after a failover to said other guest VM, and responsive to a command sent through the internal channel from the hypervisor to said other guest VM, the specified component of said other guest VM is operated as a replication sender.   
     
     
         5 . The method of  claim 4 , wherein:
 after a failover to said other guest VM, incoming data is routed to said other guest VM, and is replicated by the replication sender to storage associated with said other guest VM, and a specified component of the given guest VM is operated as a replication receiver.   
     
     
         6 . The method of  claim 1 , wherein:
 said action to correct comprises implementing failover of the specified application and the given guest VM to a second node of the HA cluster, and running the specified application on the given guest VM in the second node.   
     
     
         7 . The method of  claim 1 , wherein:
 said action to correct comprises implementing failover of the specified application to a second node of the HA cluster, and running the specified application on a guest VM other than the given guest VM, in the second node.   
     
     
         8 . The method of  claim 1 , wherein:
 the specified application running on the given guest VM is managed by an HA cluster manager associated with the hypervisor, wherein the HA cluster manager requires specified management scripts and configuration resources for its operation, and all of said scripts and configurations are contained in the hypervisor.   
     
     
         9 . The method of  claim 1 , wherein:
 the HA cluster is provided with a system for monitoring heartbeats produced by respective nodes of the HA cluster, wherein failure of a node is indicated by a deficiency of the heartbeats produced thereby.   
     
     
         10 . The method of  claim 1 , wherein:
 each node of the HA cluster has a local HA manager, wherein guest VM failures in a node are detected by its HA manager.   
     
     
         11 - 20 . (canceled)