Cluster controlling system operating on a plurality of computers in a cluster system

• US 6,088,727 A
• Filed: 10/17/1997
• Issued: 07/11/2000
• Est. Priority Date: 10/28/1996
• Status: Expired due to Fees

• Current Assignee: Mitsubishi Denki Kabushiki Kaisha (Mitsubishi Electric Corporation)
• Original Assignee: Mitsubishi Denki Kabushiki Kaisha (Mitsubishi Electric Corporation)
• Inventors: Hosokawa, Takehiko, Tsuru, Kaoru
• Primary Examiner(s): Maung, Zarni
• Application Number:
  US08/953,632
• Time in Patent Office: 998 Days
• Field of Search: 395/674, 395/676, 395/800.28, 395/200.31, 395/200.35, 395/200.43, 395/200.53, 395/200.73, 709/223, 709/201, 709/205, 709/213, 709/243, 709/225
• US Class Current: 709/223
• CPC Class Codes: G06F 11/1482 by means of middleware or OS functionality
• CPC Class Codes: G06F 11/202 where processing functionality is redundant redundant communication control functionality G06F11/2005, redundant storage control functionality G06F11/2089
• CPC Class Codes: G06F 11/3495 for systems
• CPC Class Codes: G06F 9/4856 resumption being on a different machine, e.g. task migration, virtual machine migration G06F9/5088 takes precedence
• CPC Class Codes: G06F 9/5061 Partitioning or combining of resources

Abstract

A cluster controlling system transfers packages which have been operating on one computer to another computer when a fault or failure has occurred by monitoring and controlling the packages in the entire system. When the respective packages are started-up, cluster daemons on the respective computers monitor and control resources on the operating computers. The monitored and controlled data are stored in the respective computers as local data. A manager communicates with cluster daemons on the respective computers, and stores data in a global data memory to monitor and control the entire system. The manager is actually one of the packages operating in the cluster system. If a fault or failure occurs in the manager or in the computer running the manager, the manager is re-started on another computer by a cluster daemon.

Claims

1. A cluster controlling system, comprising:

cluster daemons for communicating with a plurality of computers, and for monitoring and controlling resources;

local data memories for storing the monitoring result as local data; and

one of said plurality of computers in the cluster system including;

global data memory for storing data collected from the local data memories of the respective computers, and for being referred to from any of the computers; and

a manager for communicating with the global data memory and the cluster daemons of the respective computers, and for monitoring and controlling the overall cluster system, wherein if a fault or failure occurs in the manager or the one computer in said plurality of computers, the same manager is re-started on another computer in said plurality of computers by its corresponding cluster daemon.

2. The cluster controlling system according to claim 1, wherein said manager comprises:

a demand processing device for receiving demands from a user, and

a resource controlling device for receiving a demand from the demand processing device and referring to a resource state data base in the global data memory to transmit a resource state corresponding to the demand to a cluster daemon.

3. The cluster controlling system according to claim 2, wherein said global data memory comprises:

a resource setting file for storing process information, in response to a resource state change of one computer in the cluster system; and

wherein

said manager comprises

an automatic controlling device for controlling resources according to a definition of the resource setting file when the resource state is changed.

4. The cluster controlling system according to claim 3, wherein said resource setting file defines priority order information regarding execution and correlation-ship between packages, and wherein said automatic controlling device operates packages on respective computers according to the information defined in the resource setting file.

5. The cluster controlling system according to claim 2, wherein said manager further includes a mode managing device for managing a package according to an operation mode representing "operating", "waiting" or "testing".

6. The cluster controlling system according to claim 2, wherein said manager further comprises a log managing device for collecting logs of resource state changes which occur in the cluster system.

7. The cluster controlling system according to claim 2, wherein

said cluster daemons monitor a fault or failure and subsequent recovery of a corresponding computer, and control start-up and operations of a package;

wherein

a first computer in the plurality of the computers operates a first package; and

a second computer waits for a second package to provide the same application or service as the first package; and

wherein said manager includes

a notice processing means for receiving a resource state change notice from the cluster daemons, and

a mode controlling device for receiving the mode controlling demand from the notice processing means and then transmitting it to a cluster daemon,

wherein the manager operates a second package on a second computer by controlling the cluster daemons when a fault or failure has occurred in a first computer, and forces the first package to wait until the first computer has recovered from the failure.

8. The cluster controlling system according to claim 2, wherein

said cluster daemons monitor a fault or failure and the subsequent recovery of a corresponding computer, and control start-up and operation of a package;

wherein

a first computer in the plurality of the computers operates a first package; and

a second computer is suppressed for a second package; and

wherein

said manager includes

an output managing device for receiving a resource state change notice from the cluster daemons, and

an output suppressing device for receiving an output suppression/release demand from the output managing device and for transmitting it to a cluster daemon,

wherein the manager outputs a second package instead of a first package by controlling cluster daemons when a fault or failure occurs in a first computer operating the first package, maintains an output from the second package until a fault or failure occurs in the second computer, outputs the first package instead of the second package by controlling cluster daemons when a fault or failure occurs in the second computer operating the second package, and maintains an output from the first package until a fault or failure occurs in the first computer.

9. The cluster controlling system according to claim 2, wherein

said cluster daemons monitor fault or failure and the subsequent recovery of a corresponding computer, and control start-up and operation of the packages;

wherein

a first computer in the plurality of the computers operates a first package; and

a second computer is suppressed for a second package, and wherein

said manager includes

a package managing device for receiving and determining a type of resource state change notice from the cluster daemons, and

a package controlling device for receiving a package controlling demand and transmitting it to cluster daemon,

wherein the manager operates the second package on the second computer when a fault or failure occurs in the first computer, and places the first package in a waiting mode on a third computer.

10. The cluster controlling system according to claim 2, wherein

said cluster daemons monitor a fault or failure and the subsequent recovery of a corresponding computer, and control start-up and operation of the packages;

wherein

a first computer in the plurality of the computers operates a first package;

a second computer waits for a second package; and

wherein

said manager includes

a package managing device for receiving a monitored result from cluster daemons,

a package controlling device for receiving a package controlling demand and transmitting it to the cluster daemon,

a managing table for storing a priority order of the packages; and

wherein the manager operates the second package on the second computer when a fault or failure occurs in the first computer, retrieves a third package with lower priority than the package being operated on the first computer, stops operation of the third package, and starts-up the first package on a third computer in a waiting state.

11. The cluster controlling system according to claim 2, wherein

said cluster daemons monitor a fault or failure and the subsequent recovery of a corresponding computer, and control start-up and operation of the packages;

wherein

a first computer in the plurality of the computers operates a first group of package;

a second computer operates the same first group of packages in parallel with the first computer; and

wherein

said manager includes

a package managing device for receiving a monitored result from cluster daemons of the respective computers,

a package controlling device for receiving a package controlling demand and transmitting it to a cluster daemon,

a managing table for storing group names of the packages operating in parallel,

wherein the manager retrieves a third computer not operating the same first group of packages when a fault or failure occurs in the first computer, stops operation of the retrieved third package, and starts-up the first package on the third computer.

12. The cluster controlling system according to claim 2, wherein

said cluster daemons monitor a fault or failure and the subsequent recovery of a corresponding computer, and control start-up and operation of the packages;

wherein

a first computer in the plurality of the computers operates a higher priority of package;

a second computer operates a second higher priority of package in parallel with the first computer;

and n-th computer operates an n-th priority of package in parallel;

wherein

said manager includes

a package managing device for receiving a monitored result from the cluster daemons of the respective computers,

a package controlling device for receiving a package controlling demand and transmitting it to the cluster daemon,

a managing table for storing a priority order of respective packages on the plurality of the computers; and

wherein the manager retrieves a third computer to stop a lower priority package operating on the third computer when a fault or failure occurs in the first computer, start-up a first priority package on the third retrieved computer, re-assign resources to the respective packages based on the priority orders of the packages.

13. The cluster controlling system according to claim 2, wherein

said cluster daemons monitor a fault or failure and the subsequent recovery of a corresponding computer, and control start-up and operation of the packages;

wherein

a plurality of the computers operate respective packages in parallel, each package having a different priority; and

wherein

said manager includes

a package controlling device for receiving a package controlling demand and transmitting it to a cluster daemon,

a managing table for storing a priority order of the packages and load factors of the respective computers; and

wherein the manager retrieves a third computer to operate a higher priority package when the load factor of the first computer exceeds a predetermined load by referring to a managing table, stops operation of a package with a lower priority that is operating on the third computer, and starts-up, on the retrieved third computer, a higher priority package operating on the first computer.

14. The cluster controlling system according to claim 2, wherein

said cluster daemons monitor a fault or failure and the subsequent recovery of a corresponding computer, and control start-up and operation of the packages;

wherein

a plurality of the computers operate respective packages in parallel, each package having different kind of priority; and

wherein

said manager includes

a resource assigning managing device for receiving a resource state change notice,

a resource assignment controlling device for receiving a package controlling demand for allocating resources to the packages based on the package priority from the resource assigning managing device and transmitting it to a cluster daemon,

a managing table for storing the priority orders of the respective packages; and

wherein the manager selects a highest priority package by referring to the managing table, assigns resources to the corresponding package, and repeats this process with all lower priority packages.

15. The cluster controlling system according to claim 1, further comprising:

agents for communicating with the manager and cluster daemons.

16. The cluster controlling system according to claim 15, wherein said agent comprises:

a demand processing device for receiving resource control demands from the manager, and

a resource controlling device for receiving a demand from the demand processing device and referring to a resource state data base to transmit a resource state corresponding to the demand to a cluster daemon.

17. A cluster controlling system, comprising:

cluster daemons for communicating with packages and a plurality of computers;

agents for communicating with a corresponding cluster daemon on each of the respective computers, a global data memory and other agents;

local data memories for storing results which are monitored and controlled by the cluster daemons as local data; and

global data memory for storing data collected from the local data memories of the respective computers, and for being referred to from any of the computers via the agents, and

one computer in said plurality of computers including a manager which communicates with each of the plurality of cluster daemons and with the global data memory to control the cluster system, wherein if a fault or failure occurs in the manager or the one computer in said plurality of computers, the same manager is re-started on another computer in said plurality of computers by its corresponding cluster daemon.

18. The cluster controlling system according to claim 17, wherein said manager comprises:

a demand processing device for receiving demands from a user, and

a resource controlling device for receiving a demand from the demand processing device and referring to a resource state data base in the logbal data memory to transmit a resource state corresponding to the demand to a cluster daemon.

19. A cluster system comprising:

a plurality of computers corresponding to and communicating with a plurality of daemons;

local data memories for storing results which are monitored and controlled by the plurality of daemons;

wherein one computer in said plurality of computers further includes;

a manager which communicates with each of the plurality of daemons and with a global data memory, to control the cluster system; and

wherein if a fault or failure occurs in the manager or the one computer in said plurality of computers, the manager is re-started on another computer in said plurality of computers by its corresponding daemon.

20. A manager comprising:

a demand processing device for receiving demands from a user;

a resource controlling device which receives a demand from the demand processing device, refers to a database in a global memory, and transmits a resource state corresponding to said demand to a first daemon of a first computer which controls resources,

wherein if a fault or failure occurs in the manager or in the first computer operating the manager, the manager is re-started on a second computer by the second computer'"'"'s corresponding daemon.