Distributed processing system with checkpoint restart facilities wherein checkpoint data is updated only if all processors were able to collect new checkpoint data

Claim 2

A distributed processing system with a restart processing function comprising: a first data processor and a second data processor, each executing a distributed processing; communication control means, installed in each said first processor and said second data processor, for allowing transmission and reception of data between said first data processor and said second data processor; checkpoint processing means, installed in each said first data processor and said second data processor, for determining new checkpoint data as current checkpoint data for executing restart processing when an ACK signal is transmitted from said second data processor to said first data processor in response to a request to collect said new checkpoint data made by said first data processor, and determining old checkpoint data as said current checkpoint data when no ACK signal is transmitted from said second data processor in response to said request made by said first data processor; data storage means, installed in each said first data processor and said second data processor, for storing the current checkpoint data determined by said checkpoint processing means; and restart processing means, installed in each said first data processor and said second data processor, for restarting execution of said distributed processing in accordance with said current checkpoint data stored in said data storage means.

Claim 3

A system according to claim 1, further comprising failure detecting means, installed in each said first data processor and said second data processor, for detecting a failure in at least one of said first data processor and said second data processor, said restart processing means responding to detection of a failure by said failure detecting means by executing said restart processing based on said current checkpoint data stored in said data storage means.

Claim 4

A distributed processing system with a restart processing function comprising: a first data processor and a second data processor, each executing a distributed processing; communication control means, installed in each said first data processor and said second data processor, for allowing transmission and reception of data between said first data processor and said second data processor; first checkpoint processing means, installed in each said first data processor and said second data processor, for generating old checkpoint data and new checkpoint data necessary for restart processing in accordance with a point at which said distributed processing is being executed; data storage means, installed in each said first data processor and said second data processor and having storage areas, for storing said old checkpoint data and said new checkpoint data generated by said first checkpoint processing mans; second checkpoint processing means for determining said new checkpoint data as current checkpoint data when an ACK signal is transmitted from said second data processor to said first data processor in response to a request to collect said new checkpoint data made by said first data processor, and determining said old checkpoint data as said current checkpoint data when no ACK signal is transmitted from said second data processor in response to said request made by said first data processor; and restart processing means, installed in each said first data processor and said second data processor, for restarting execution of said distributed processing in accordance with said current checkpoint data stored in said data storage means.

Claim 5

A distributed processing system with a restart processing function comprising: a first data processor and a second data processor, each executing a distributed processing; communication control means, installed in each said first data processor and said second data processor, for allowing transmission and reception of data between said first data processor and said second data processor; first checkpoint processing means, installed in each said first data processor and said second data processor, for generating old checkpoint data and new checkpoint data necessary for restart processing in accordance with a point at which said distributed processing is being executed; data storage means, installed in each said first data processor and said second data processor and having storage areas, for storing said old checkpoint data and said new checkpoint data generated by said first checkpoint processing mans; second checkpoint processing means for determining said new checkpoint data as current checkpoint data when an ACK signal is transmitted from said second data processor to said first data processor in response to a request to collect said new checkpoint data made by said first data processor, and determining said old checkpoint data as said current checkpoint data when no ACK signal is transmitted from said second data processor in response to said request made by said first data processor; and restart processing means, installed in each said first data processor and said second data processor, for restarting execution of said distributed processing in accordance with said current checkpoint data stored in said data storage means.

Claim 6

A system according to claim 3, further comprising failure detecting means installed in each said first data processor and said second data processor for detecting a failure in at least one of said first data processor and said second data processor, said restart processing means responding to detection of a failure by said failure detecting means to execute said restart processing based on said current checkpoint data stored in said data storage means.

Claim 7

A distributed processing system comprising: a plurality of data processing apparatuses; communication control means, installed in each of said data processing apparatuses, for controlling transmission and reception of data between said data processing apparatuses; data storage means, installed in each of said data processing apparatuses, for storing new and old checkpoint data; checkpoint processing means, installed in each of said data processing apparatuses, for storing first new checkpoint data in said data storage means for one of said data processing apparatuses and requesting that another of said data processing apparatuses collect second new checkpoint data in response to transferred data when said communication control means transfers data to said another of said data processing apparatuses, and for clearing old checkpoint data in response to an acknowledge signal representing that said another of said data processing apparatuses collected said second new checkpoint data accurately; and restart processing means, installed in each of said data processing apparatuses, for reading one of said first new checkpoint data and said old checkpoint data from said data storage means and restarting processing based on read checkpoint data when said checkpoint processing means does not receive said acknowledge signal.

Claim 8

A distributed processing system comprising: a plurality of data processing apparatuses; communication control means, installed in each of said data processing apparatuses, for controlling transmission and reception of data between said data processing apparatuses; data storage means, installed in each of said data processing apparatuses, for storing new and old checkpoint data; checkpoint processing means, installed in each of said data processing apparatuses, for storing first new checkpoint data in said data storage means for one of said data processing apparatuses and requesting that another of said data processing apparatuses collect second new checkpoint data in response to transferred data when said communication control means transfers data to said another of said data processing apparatuses, and for clearing old checkpoint data in response to an acknowledge signal representing that said another of said data processing apparatuses collected said second new checkpoint data accurately; and restart processing means, installed in each of said data processing apparatuses, for reading one of said first new checkpoint data and said old checkpoint data from said data storage means and restarting processing based on read checkpoint data when said checkpoint processing means does not receive said acknowledge signal.

Claim 9

A system according to claim 5, in which said restart processing means reads said first new checkpoint data and requests said another of said data processing apparatuses to read said second new checkpoint data when said checkpoint processing means does not receive said acknowledge signal, and restarts processing based on said first new checkpoint data stored in said data storage means in response to another acknowledge signal representing that said another data processing apparatus collected said second new checkpoint data accurately.