Kabushiki Kaisha Toshiba

Toshiba Solutions Corporation

Hideaki Hirayama

Toshio Shirakihara

Tatsunori Kanai

Kiyoko Sato

Kuniyasu Shimizu

Hiroshi Sakai

Shigefumi Ohmori

Tetsuya Iinuma

Kuninori Tanaka

Hideki Yoshida

Makoto Honda

Takahiro Fujii

Takeo Takahashi

Tetsuro Kimura

Yoshio Masubuchi

Nobuo Sakiyama

Satoshi Hoshina

Kenichi Mizoguchi

Masaharu Nozaki

Tohru Iijima

Yuuichi Koba

Nobuyoshi Kamei

Teruyoshi Zemmyo

Tomofumi Shimada

A server borrowing apparatus, includes a storing unit that stores lender installation setting information which includes a setting item determined by a server lender for installation of a server which is lent out as resource, a receiving unit that receives a request to lend out the server as the resource and borrower installation setting information which includes a setting item determined by a server borrower for the installation of the server, from a borrowing control apparatus which is connected to a network to manage the server and to control server borrowing, a combining unit that combines the setting item in the borrower installation setting information and the setting item in the lender installation setting information to generate installation setting information for the installation of the server at the server borrowing, and an installing unit that performs the installation of the server based on the installation setting information.

Server borrowing apparatus allocating server to user group and computer product therefor

Description of a parallel program is facilitated in a cluster system which loosely couples a plurality of server computers each having a cluster file system. In the cluster system, an application program is executed to generate a process. Using the cluster file system, the process maps a file recorded in a shared disk device to a cluster shared memory (distributed shared memory) area virtually provided in an address space arranged on main memory. By doing so, an access to files is processed as an access to main memory.

Cluster system, memory access control method, and recording medium

Methods and systems consistent with this invention link a program module, which implements dynamic linkage to an external function and executes the external function. Such methods and systems replace a function name described in a header portion of a program module and a function name to be referred to by the program module with other names, and link functions having the same names as the replaced function names and executing the linked functions in place of an original function.

Method of changing a dynamic link library function efficiently and a computer system for executing the same

A commit controlling scheme for a transaction processing in a system utilizing the check point/roll back scheme, which is capable of guaranteeing the reliability of data of the transaction processing even when the transaction processing and the check point/roll back scheme are employed simultaneously, is described. An issuance of an external output for notifying to a user that a transaction has been committed is delayed until after a new check point for the transaction processing process is generated by the fault recovery mechanism at a timing later than a commit of the transaction by the transaction processing process. This can be realized by delaying an issuance of a transaction commit notice from the transaction processing process, or by intercepting a transaction commit notice sent from the transaction processing process, or else by delaying an issuance of the external output by the client process after receiving a transaction commit notice from the transaction processing process.

Commit controlling scheme for transaction processing in system utilizing check point/roll back scheme

When the item appearance frequency count up section counts up the appearance frequency of an item in a shared memory space on the basis of transaction data, a count up log transfer section receives the count up log recorded by a count up log recording section and transfers it to other nodes. A count up log receiving section transfers the count up log transferred from each of the remaining nodes to a count up log reflecting section. The count up log reflecting section, which has received this count up log, reflects its contents in the appearance frequency of the corresponding item in the shared memory space.

Distributed shared memory system and method of controlling distributed shared memory

A process state management scheme capable of acquiring process states consistently even in a case where a new process is generated from some process, while using the synchronous checkpointing method. This scheme prohibits a new process generation during a process state acquisition, which can be realized by judging whether a process generation request by a first process for generating a second process is prior to a process state acquisition request or not, and generating the second process from the first process accordingly. This scheme also prohibits a process state acquisition during a new process generation, which can be realized by acquiring a process state at each of the first and second processes after a notice of the identifier of the second process from the first process is received, when a notice of the identifier of the second process from the second process is prior to the notice from the first process.

Method and system for process state management using checkpoints

A scheme for restarting processes at distributed checkpoints in a client-server computer system, in which a fault in one client computer does not affect the server computer and the other client computers. In this scheme, a fault occurring in one computer among of a plurality of computers constituting a client-server computer system is detected while these plurality of computers are executing respective processes, and whether that one computer in which the fault is detected is a server computer or not is judged. Then, related processes executed on these plurality of computers are restarted when that one computer is judged as the server computer, whereas no process executed on these plurality of computers is restarted when that one computer is not judged as the server computer. It is also possible to specify a restart information for each process indicating how processes should be restarted when a process fault occurs in each process, and to restart selected processes executed on these plurality of computers according to the restart information for a process in which the process fault is detected.

Scheme for restarting processes at distributed checkpoints in

A method for recovering from software fault in a fault tolerant computing system includes a system status recording step to record the system status at the occurrence of the above software fault when the above fault is judged to be a software fault by a fault identifying step, a software fault factor diagnosing step to diagnose the fault factor of the above software fault, a software fault recovery action determining step to determine a recovery action to the above fault factor of the above software fault, and a software fault recovery action executing step to execute the recovery action the above fault factor of the above software fault determined by the above software fault recovery action determining step after roll back.

Method and apparatus for recovering from software faults

The distributed memory type multiprocessor computing system according to the present invention includes a plurality of nodes which are coupled with each other so as to correspond to a tree construction by communicating switches. Each of nodes acquires checkpoints at a predetermined time for reexecuting the respective data processing from a previously acquired checkpoint when a fault occurred. Each of nodes starts the checkpoint acquisition in synchronous with all another nodes by prohibiting data transfer through the communication switch to another nodes and by making completion of data reception through the communication switch for keeping vacant state which no data exist in the communication switch.

Method and apparatus for recovering from faults in distributed memory

A checkpoint restart type computer system which executes a program upon taking a checkpoint periodically for rolling back to a prior checkpoint if an error status is detected during an execution of a program, including a memory for storing a program which includes a plurality of ESD steps between particular two checkpoints, and a processor for executing the program with holding a checkpoint periodically. The processor includes a memory for maintaining a checkpoint image which has been taken during execution of the program. The ESD steps in the program contain at least a designated execution level for the ESD step and an error status detecting code. The computer system further includes a designator for designating the execution level for the program and a classifier for classifying the execution of the ESD steps by comparing the designated execution level and the allowed execution level for the ESD sequence. When an error status is detected during the execution of the program, the designator changes the execution level of the program from a low to a high frequency so as to increase the number of executions of the ESD steps for specifying a particular position of an error status as early as possible.

AC checkpoint restart type fault tolerant computer system

A computer system which can achieve rollback operation when a fault occurs in the system without waiting for side-tracking of pre-update data during updating of a file. When a file write request has been made, 'file writing information' pertaining to the file write is saved in a pending queue and only a primary file is immediately updated. After a checkpoint has been acquired, the 'file writing information' saved in the pending queue is shifted to a confirmed queue, and is then written to a back-up file. When performing recovery, all pre-update data which corresponds to the data which has been updated since the last checkpoint acquired is read from the back-up file, based on the 'file writing information' saved in the pending queue. The primary file is then restored to its state at the checkpoint time by using the pre-update data which has been read from the backup file.

Checkpointing computer system having duplicated files for executing

The present invention provides a computer system with an interruption resistance which construction costs are greatly reduced by eliminating lock-run-out sequence when restarting is performed. In a computer system with an interruption resistance, comprising a check point restart function, special process for taking check points are respectively provided for processors included in the computer system. When taking a check point, the special process for taking check points is brought into an ready state by a wake-up portion. After the special process for taking check points takes a check point, the special process for taking check points is brought into a sleep state again. In this manner, check points are not taken when executing an arbitrary process, and therefore, lock-run-out can be eliminated when restarting is performed.

Computer system having a checkpoint and restart function

A computer system including a plurality of processors and a copyback cache memory and a method for periodically obtaining a first and second phase memory checkpoints and recovering from faults using the checkpoints, are disclosed. The computer system has cache flush hardware for executing a cache flush operation independently of the processors, including cache flush starters for starting the cache flush hardware, and cache flush end detectors for detecting the end of the cache flush operation. During a first phase checkpoint, the cache flush and normal data processing are done in parallel. When the cache flush end detectors detect the termination of the cache flush hardware, the processors suspend normal data processing, and a second checkpoint step is executed in which the processors save the context of the processors in the main memory and the cache flush hardware is invoked again so that dirty data in the cache memory is written back in to the main memory.

Computer system and method for obtaining memory check points and

A checkpoint based communication processing system and method in which an operation is executed and during execution checkpoints are periodically taken in order to recover from a failure which may be detected in the operation between any two checkpoints. Communication processing is controlled in accordance with a send-to-request instruction from an application. A plurality of communication packets for the communication processing are generated and a sequence numbered for a checkpoint is incremented each time when a checkpoint has periodically been taken during the operation. Affixed to each communication packet is attribute information in the form of the checkpoint sequence number existing at the time of generation of the communication packet, with plural communication packets generated between incrementing of the sequence number having affixed thereto the same sequence number. Whether or not the communication is promptly transferred is determined based on a comparison of the fixed attribute information and the latest sequence number. In another embodiment, the attribute information indicates whether or not the packet is the last of the generated of the plurality of transfer packets, and transfer packets fixed with attribute information which does not indicate the last position is transferred immediately, but the transfer packet which has affixed attribute information indicating the last position is delayed in transmission until a next checkpoint has been taken.

Method and apparatus for a checkpoint based communication processing

Log memories for recording updated history of a main memory are provided. CPUs record the updated history of the main memory to either of the log memories and writes context thereof and content of a cache memory to the main memory at a checkpoint acquisition. The updated history of the main memory is switched from one of CPUs that has finished a checkpoint processing to other one of the log memories in which the CPUs do not use to record the updated history of the main memory. Normal processing is restarted without waiting for finishing the checkpoint acquisition of the other ones of CPUs.

Growing community of inventors

Kanagawa-ken, Japan

Hideaki Hirayama