Method and apparatus for automatic recovery from a failed node concurrent maintenance operation

A method in a data processing system for automatic recovery from a failed node concurrent maintenance remove operation, the method comprising:

• removing resources on a processor node to be removed from a plurality of processor nodes in the data processing system;
• disabling communication links between the plurality of processor nodes;
• updating values of a current mode register, wherein the values comprise configuration settings of processors of the plurality of processor nodes excluding the processor node to be removed;
• sending a test command to remaining processors of the plurality of processor nodes;
• detecting one of a response from the remaining processors of the plurality of processor nodes and a timeout;
• and
• responsive to detecting a response from the remaining processors, determining if the response is a correct response matching the test command if the response is a correct response, initializing communications between the plurality of processor nodes in the data processing system excluding the processor node to be removed;
• and responsive to one of detecting the timeout and determining that the response is not a correct response, check stopping the data processing system.

A data processing system for automatic recovery from a failed node concurrent maintenance operation, the data processing system comprising:

• a processor;
• a memory connected to a processor via a processor bus;
• wherein the processor is adapted to execute computer implemented instructions in the memory to;
• disable communications between a plurality of processor nodes in the data processing system;
• send a first test command to processors of a new processor node to be added to the data processing system;
• detect one of a response from the processors of the new processor node and a timeout;
• wherein the processor is further adapted to execute computer implemented instructions in the memory to:
• determine if the response is a correct response matching the first test command,
• responsive to detecting the response from the processors of the new processor node;
• update values of a current mode register if the response is a correct response, wherein the values comprise configuration settings of processors of the new processor node;
• update all processors in the plurality of processor nodes to use values of the current mode register
• abort an add operation of the new processor node, responsive to one of detecting a timeout and determining that the response is not a correct response;
• send a system attention to one of a service processor and a system software of the data processing system;
• send a second test command to all processors in the plurality of processor nodes;
• detect one of a response from all processors in the plurality of processors and the timeout;
• determining if the response is a correct response matching the second test command;
• and
• initialize communications between the plurality of processor nodes in the data processing system including the new processor node, if the response is a correct response.