Apparatus and method for detecting the reset of a node in a cluster computer system

A method for managing a cluster of nodes, said method comprising the steps of:

• coupling first and second nodes in said cluster;
• said first node recognizing said second node as an operational node in said cluster;
• subsequently, said second node being decoupled from said cluster;
• subsequently, said second node recoupling with said cluster;
• subsequently, said first node recognizing that said second node has been recoupled with said cluster;
• in response to the subsequent recognizing step, said first node diagnosing at least one characteristic of said second node relating to operational status of said second node;
• and
• subsequently, if said second node is operational, said second node verifying its configuration to said first node and then said first node sending a command to be executed by said second node for load balancing;
• and wherein
• immediately before said second node failed, said second node was a master node and said first node was a slave node, and after said second node becomes operational after said diagnosing step, said first node is a master node and said second node is a slave node.

A method for managing a cluster of nodes, said method comprising the steps of:

• coupling first and second nodes in said cluster;
• said first node recognizing said second node as an operational node in said cluster;
• subsequently, said second node being decoupled from said cluster;
• subsequently, said second node recoupling with said cluster;
• subsequently, said first node recognizing that said second node has been recoupled with said cluster;
• in response to the subsequent recognizing step, said first node diagnosing at least one characteristic of said second node relating to operational status of said second node;
• and
• subsequently, if said second node is operational, said second node verifying its configuration to said first node and then said first node sending a command to be executed by said second node for load balancing;
• and wherein
• said diagnosing step comprises the step of checking status of a cable from said second node into said cluster.

A method for managing a cluster of nodes, said method comprising the steps of:

• coupling first and second nodes in said cluster;
• said first node recognizing said second node as an operational node in said cluster;
• subsequently, said second node being decoupled from said cluster;
• subsequently, said second node recoupling with said cluster;
• subsequently, said first node recognizing that said second node has been recoupled with said cluster;
• in response to the subsequent recognizing step, said first node diagnosing at least one characteristic of said second node relating to operational status of said second node;
• and
• subsequently, if said second node is operational, said second node verifying its configuration to said first node and then said first node sending a command to be executed by said second node for load balancing;
• and wherein
• said first and second nodes are coupled to each other via a bus;
• and
• said step of said first node recognizing that said second node has been recoupled with said cluster comprises the step of said first node recognizing a resetting of said bus.

A method for managing a cluster of nodes, said method comprising the steps of:

• coupling first and second nodes in said cluster;
• said first node recognizing said second node as an operational node in said cluster;
• subsequently, said second node being decoupled from said cluster;
• subsequently, said second node recoupling with said cluster;
• subsequently, said first node recognizing that said second node has been recoupled with said cluster;
• in response to the subsequent recognizing step, said first node diagnosing at least one characteristic of said second node relating to operational status of said second node;
• and
• subsequently, if said second node is operational, said second node verifying its configuration to said first node and then said first node sending a command to be executed by said second node for load balancing;
• and wherein before the verifying step, the following step is performed:
• said first node determines whether said second node failed between said coupling and recoupling of said second node with said cluster.

A method for managing a cluster of nodes, said method comprising the steps of:

• coupling first and second nodes in said cluster;
• said first node recognizing said second node as an operational node in said cluster;
• subsequently, said second node being decoupled from said cluster;
• subsequently, said second node recoupling with said cluster;
• subsequently, said first node recognizing that said second node has been recoupled with said cluster;
• in response to the subsequent recognizing step, said first node diagnosing at least one characteristic of said second node relating to operational status of said second node;
• and
• subsequently, if said second node is operational, said second node verifying its configuration to said first node and then said first node sending a command to be executed by said second node for load balancing;
• and wherein after said verifying step, said second node becomes a slave node.

A computer-readable medium for data storage wherein is located a computer program for causing a first node in a computer system, said first node having a first bus controller, to detect failure of a second node in said computer system, said second node coupled to said first node by means of a bus and a second bus controller, by

• firstly recognizing on said first node said second node as either a master or a slave node in said cluster;
• then failing to keep a node communicatively coupled to said cluster;
• then communicatively coupling said failed node to said cluster a second time, including resetting said bus;
• secondly recognizing, on said first node, said second coupling, including recognizing said reset of said bus;
• in response to said second recognition, querying, by means of said first node, each node of said cluster for failure-status information;
• determining, on said first node and by means of said failure-status information, whether said second node failed between said first and second couplings;
• ceasing to recognize said second node as said master node, negotiating slave membership in said cluster for said second node's
• and thirdly recognizing said first node as said master node, on determining on said first node that said second node failed between said first and second couplings and that said second node was secondly recognized as said master node;
• negotiating slave membership in said cluster for said second node's on determining on said first node that said second node failed between said first and second couplings and that said second node was secondly recognized as a slave node;
• and
• accepting said second node as a member of said cluster with its master/slave status intact on determining on said first node that said second node did not fail between said first and second couplings.

A first computer node in a cluster for managing a second computer node in said cluster, said first and second nodes being coupled by a bus, said first node comprising:

• means for recognizing that said second node is operational and coupled to said cluster;
• means for recognizing that said second node has been recoupled to said cluster after said second node became decoupled from said cluster;
• means for diagnosing at least one characteristic of said second node relating to operational status of said second node after said second node is recoupled to said cluster;
• means, active if said second node is diagnosed as operational, for receiving a verification of configuration from said second node;
• and
• means, after said verification, for sending to said second node a command to be executed by said second node;
• and wherein said first and second nodes are both coupled to a same plurality of RAID storage controllers, and said command is a storage access command.

A first computer node in a cluster for managing a second computer node in said cluster, said first and second nodes being coupled by a bus, said first node comprising:

• means for recognizing that said second node is operational and coupled to said cluster;
• means for recognizing that said second node has been recoupled to said cluster after said second node became decoupled from said cluster;
• means for diagnosing at least one characteristic of said second node relating to operational status of said second node after said second node is recoupled to said cluster;
• means, active if said second node is diagnosed as operational, for receiving a verification of configuration from said second node;
• and
• means, after said verification, for sending to said second node a command to be executed by said second node;
• and wherein immediately before said second node was decoupled from said cluster, said second node was a master node and said first node was a slave node, and after said second node is recoupled to said cluster and is operational, said first node is a master node and said second node is a slave node.

A first computer node in a cluster for managing a second computer node in said cluster, said first and second nodes being coupled by a bus, said first node comprising:

• means for recognizing that said second node is operational and coupled to said cluster;
• means for recognizing that said second node has been recoupled to said cluster after said second node became decoupled from said cluster;
• means for diagnosing at least one characteristic of said second node relating to operational status of said second node after said second node is recoupled to said cluster;
• means, active if said second node is diagnosed as operational, for receiving a verification of configuration from said second node;
• and
• means, after said verification, for sending to said second node a command to be executed by said second node;
• and wherein the diagnosing means comprises means for checking a cable from said second node into said cluster.

A first computer node in a cluster for managing a second computer node in said cluster, said first and second nodes being coupled by a bus, said first node comprising:

• means for recognizing that said second node is operational and coupled to said cluster;
• means for recognizing that said second node has been recoupled to said cluster after said second node became decoupled from said cluster;
• means for diagnosing at least one characteristic of said second node relating to operational status of said second node after said second node is recoupled to said cluster;
• means, active if said second node is diagnosed as operational, for receiving a verification of configuration from said second node;
• and
• means, after said verification, for sending to said second node a command to be executed by said second node;
• and wherein said means for recognizing that said second node has been recoupled to said cluster after said second node became decoupled from said cluster comprises means for recognizing a resetting of a bus.

A first computer node in a cluster for managing a second computer node in said cluster, said first and second nodes being coupled by a bus, said first node comprising:

• means for recognizing that said second node is operational and coupled to said cluster;
• means for recognizing that said second node has been recoupled to said cluster after said second node became decoupled from said cluster;
• means for diagnosing at least one characteristic of said second node relating to operational status of said second node after said second node is recoupled to said cluster;
• means, active if said second node is diagnosed as operational, for receiving a verification of configuration from said second node;
• means, after said verification, for sending to said second node a command to be executed by said second node;
• and
• means for determining whether said second node failed between said coupling and recoupling of said second node with said cluster.

A computer program product for execution in a first computer node in a cluster for managing a second computer node in said cluster, said first and second nodes being coupled by a bus, said computer program product comprising:

• a computer readable medium;
• first program instructions for recognizing that said second node is operational and coupled to said cluster;
• second program instructions for recognizing that said second node has been recoupled to said cluster after said second node became decoupled from said cluster;
• third program instructions for diagnosing at least one characteristic of said second node relating to operational status of said second node after said second node is recoupled to said cluster;
• fourth program instructions, active if said second node is diagnosed as operational, for receiving a verification of configuration of said second node;
• and
• fifth program instructions, after said verification, for sending to said second node a command to be executed by said second node;
• and wherein
• all of said program instructions are recorded on said medium;
• and said first and second nodes are both coupled to a same plurality of RAID storage controllers, and said command is a storage access command.

A computer program product for execution in a first computer node in a cluster for managing a second computer node in said cluster, said first and second nodes being coupled by a bus, said computer program product comprising:

• a computer readable medium;
• first program instructions for recognizing that said second node is operational and coupled to said cluster;
• second program instructions for recognizing that said second node has been recoupled to said cluster after said second node became decoupled from said cluster;
• third program instructions for diagnosing at least one characteristic of said second node relating to operational status of said second node after said second node is recoupled to said cluster;
• fourth program instructions, active if said second node is diagnosed as operational, for receiving a verification of configuration of said second node;
• and
• fifth program instructions, after said verification, for sending to said second node a command to be executed by said second node;
• and wherein
• all of said program instructions are recorded on said medium;
• and
• immediately before said second node was decoupled from said cluster, said second node was a master node and said first node was a slave node, and after said second node is recoupled to said cluster and becomes operational, said first node is a master node and said second node is a slave node.