Failover method in a cluster computer system

A failover method in a cluster computer system comprising:

• a plurality of computers, with respect to the execution of a certain application, at least one of the computers being used as a computer of an active system, and at least one of the other computers being used as a computer of a standby system that takes over processing performed in the computer of the active system;
• a heartbeat path used by each of the plurality of computers to monitor states of the other computers;
• a reset path used by each of the plurality of computers to reset any of the other computers;
• and
• a reset control unit for controlling reset operation,
• said failover method comprising the steps of:
• providing each of the plurality of computers with the priority by which said each of the plurality of computers resets any of the other computers;
• issuing a reset command for resetting another computer as a target on the basis of monitoring the states of the other computers through the heartbeat path, a reset target of said reset command being any of the other computers;
• comparing the priority of a source computer of a reset command, which has been newly inputted into the reset control unit, with the priority of a destination computer of the newly inputted reset command;
• in said comparison step, if the priority of the source computer is higher than that of the destination computer, transmitting the newly inputted reset command to the destination computer;
• in said comparison step, if the priority of the source computer is not higher than that of the destination computer, bringing the newly inputted reset command into a waiting state in the reset control unit, and after a lapse of a specified period of wait time, transmitting, to the destination computer, the newly inputted reset command that has been brought into a waiting state;
• and
• upon completion of reset operation performed by the transmitted reset command, deleting, from among reset commands that are in a waiting state in the reset control unit, a reset command conflicting with the completed reset operation from the reset control unit.

A failover method in a cluster computer system comprising:

• a plurality of computers, with respect to the execution of a certain application, at least one of the computers being used as a computer of an active system, and at least one of the other computers being used as a computer of a standby system that takes over processing performed in the computer of the active system;
• a heartbeat path used by each of the plurality of computers to monitor states of the other computers;
• a reset path used by each of the plurality of computers to reset any of the other computers;
• and
• a reset control unit for controlling reset operation,
• said failover method comprising the steps of:
• providing each of the plurality of computers with the priority by which said each of the plurality of computers resets any of the other computers;
• issuing a reset command for resetting another computer as a target on the basis of monitoring the states of the other computers through the heartbeat path, a reset target of said reset command being any of the other computers;
• comparing the priority of a source computer of a reset command, which has been newly inputted into the reset control unit, with the priority of a destination computer of the newly inputted reset command;
• in said comparison step, if the priority of the source computer is higher than that of the destination computer, judging whether or not the newly inputted reset command conflicts with a reset command that is currently being executed, and if it is judged that the newly inputted reset command conflicts with no reset command, transmitting the newly inputted reset command to the destination computer;
• as a result of the judgment, if it is judged that the newly inputted reset command conflicts with a reset command that is currently being executed, bringing the newly inputted reset command into a waiting state in the reset control unit, and after a lapse of a specified period of time, judging again whether or not the reset command which has been brought into a waiting state conflicts with a reset command that is currently being executed;
• in said comparison step, if the priority of the source computer is not higher than that of the destination computer, bringing the newly inputted reset command into a waiting state in the reset control unit, and after a lapse of a specified period of time, transmitting, to the destination computer, the reset command that has been brought into a waiting state;
• and
• upon completion of reset operation performed by the transmitted reset command, deleting, from among reset commands that are in a waiting state in the reset control unit, a reset command conflicting with the completed reset operation from the reset control unit.

A failover method in a cluster computer system comprising:

• a plurality of computers, with respect to the execution of a certain application, at least one of the computers being used as a computer of an active system, and at least one of the other computers being used as a computer of a standby system that takes over processing performed in the computer of the active system;
• a heartbeat path used by each of the plurality of computers to monitor states of the other computers;
• a reset path used by each of the plurality of computers to reset any of the other computers;
• and
• a reset control unit for controlling reset operation,
• said failover method comprising the steps of:
• storing beforehand, in the reset control unit, a group of computers to which each of the plurality of computers belongs;
• issuing a reset command for resetting another computer as a target on the basis of monitoring the states of the other computers through the heartbeat path, a reset target of said reset command being any of the other computers;
• and
• comparing a group to which a source computer of an issued reset command belongs with a group to which a destination computer of the issued reset command belongs, and if both of the groups agree with each other, transmitting the issued reset command to the destination computer, whereas if both of the groups do not agree with each other, notifying the source computer that the reset command cannot be executed.

A failover method in a cluster computer system comprising:

• a plurality of computers, with respect to the execution of a certain application, at least one of the computers being used as a computer of an active system, and at least one of the other computers being used as a computer of a standby system that takes over processing performed in the computer of the active system;
• a heartbeat path used by each of the plurality of computers to monitor states of the other computers;
• a reset path used by each of the plurality of computers to reset any of the other computers;
• and
• a reset control unit for controlling reset operation,
• said failover method comprising the steps of:
• storing beforehand, in the reset control unit, the priority of each of the plurality of computers with respect to reset operation, said each computer resetting any of the other computers;
• issuing a reset command for resetting another computer as a target on the basis of monitoring the states of the other computers through the heartbeat path, a reset target of said reset command being any of the other computers;
• and
• comparing the priority of a source computer of an issued reset command with the priority of a destination computer of the issued reset command, and if the priority of a source computer of a newly inputted reset command is higher than the priority of a destination computer of the newly inputted reset command, transmitting the issued reset command to the destination computer, whereas if the priority of a source computer of a newly inputted reset command is lower than the priority of a destination computer of the newly inputted reset command, notifying the source computer that the reset command cannot be executed.

A computer system having a cluster configuration, said computer system comprising:

• a plurality of computers, with respect to the execution of a certain application, at least one of the computers being used as a computer of an active system, and at least one of the other computers being used as a computer of a standby system that takes over processing performed in the computer of the active system;
• a heartbeat path used by each of the plurality of computers to monitor states of the other computers;
• and
• a reset control unit for receiving a reset command, which is issued by each of the plurality of computers to any of the other computers, making a judgment, and on the basis of the result of the judgment, transmitting the received reset command to a target computer to be reset, wherein:
• said reset control unit includes:
• a buffer for storing information about the reset command that has been transmitted to the target computer to be reset;
• and
• a control module for controlling the buffer, and comparing the inputted reset command with the information stored in the buffer so as to make said judgment.

A computer system having a cluster configuration, said computer system comprising:

• a plurality of computers, with respect to the execution of a certain application, at least one of the computers being used as a computer of an active system, and at least one of the other computers being used as a computer of a standby system that takes over processing performed in the active system;
• a heartbeat path used by each of the plurality of computers to monitor states of the other computers;
• and
• a reset control unit for receiving a reset command, which is issued by each of the plurality of computers to any of the other computers, making a judgment, and on the basis of the result of the judgment, transmitting the received reset command to a target computer to be reset, wherein:
• said reset control unit includes:
• a first buffer for storing information about the reset command that has been transmitted to the target computer to be reset;
• a second buffer for storing a reset command that is brought into a waiting state as a result of postponing transmission of the reset command to the target computer to be reset;
• and
• a control module for comparing the inputted reset command with the information stored in the buffers so as to make said judgment, and for, if it is judged that the inputted reset command conflicts with a reset command that is being executed, postponing transmission of the inputted reset command to the target computer to be reset so that the received reset command is brought into a waiting state in the second buffer.

A computer system having a cluster configuration, said computer system comprising:

• a plurality of computers, with respect to the execution of a certain application, at least one of the computers being used as a computer of an active system, and at least one of the other computers being used as a computer of a standby system that takes over processing performed in the active system;
• a heartbeat path used by each of the plurality of computers to monitor states of the other computers;
• and
• a reset control unit for receiving a reset command, which is issued by each of the plurality of computers to any of the other computers, making a judgment, and on the basis of the result of the judgment, transmitting the received reset command to a target computer to be reset, wherein:
• said reset control unit includes:
• a first buffer for storing:
• information about the reset command that has been transmitted to the target computer to be reset, and that is currently being executed;
• and the priority by which each of the plurality of computers resets any of the other computers as a result of monitoring states of the other computers;
• a second buffer for storing a reset command that is brought into a waiting state as a result of postponing transmission of the reset command to the target computer to be reset;
• and
• a control module for comparing the priority of a source of an inputted reset command with the priority of a destination of the inputted reset command, and if the priority of a destination of the inputted reset command is higher than that of a source of the inputted reset command, bringing the inputted reset command into a waiting state in the second buffer so as to wait for a reset command from a computer whose priority is higher, whereas if the priority of a destination of the inputted reset command is lower than that of a source of the inputted reset command, making a further judgment as to whether or not the inputted reset command conflicts with a reset command that is being executed, and, if it is judged that a reset conflict occurs, bringing the inputted reset command into a waiting state in the second buffer so as to wait for the completion of the reset command that is being executed, whereas if it is judged that no reset conflict occur, transmitting the inputted reset command to the target computer to be reset.