System and method for preventing data corruption in computer system clusters

A system comprising:

• a first node comprising a first processor and a memory coupled to the first processor;
• a protection module configured to receive an indication that a distributed computer system has partitioned into a first subcluster and a second subcluster and to take control of a majority of at least one coordinator storage device, wherein the first subcluster comprises the first node, and wherein the first node comprises the protection module;
• a configuration module operable to configure the at least one coordinator storage device;
• and
• an administration module operable to send one or more of a plurality of commands to the at least one coordinator storage device, wherein
• the protection module is configured to cause the administration module to send a command to unregister a key to the at least one coordinator storage device, wherein the key is associated with a node identified for an attempt to unregister, and
• the protection module is further configured to cause the administration module to send an unregister with resource command to a shared data resource in response to the protection module having unregistered the key on a majority of the at least one coordinator storage device.

A computer readable storage medium comprising program instructions for transporting data in a distributed computing system including at least one coordinator storage device, wherein the program instructions execute on a first node of the distributed computing system and wherein the program instructions are executable to:

• detect when the distributed computing system is partitioned into a first subcluster comprising the first node and a second subcluster;
• identify, for an attempt to unregister, a first node of the second subcluster;
• attempt to unregister a key associated with the first node of the second subcluster on a majority of the at least one coordinator storage device;
• send an unregister with resource command to a shared data resource in response to having unregistered the key on a majority of the at least one coordinator storage device;
• and
• remove the first node from the distributed computing system when the attempting is unsuccessful.

An apparatus comprising:

• a network interface means for accessing a network comprising a first node and a second node;
• a resource fencing means for detecting partitioning of a computer system cluster into a first subcluster comprising the first node and a second subcluster comprising the second node and preventing data corruption of a shared data resource;
• a resource commanding means for sending at least one command to the shared resource and at least one command to unregister a key to at least one coordinator storage device, wherein the key is associated with a node identified for unregistering;
• a protection means for sending an unregister with resource command to the shared data resource in response to having unregistered the key on a majority of the at least one coordinator storage device;
• and
• a node commanding means for ejecting from the computer system cluster the first node when the resource fencing means is unable to control the majority of the at least one coordinator resource.