Method and apparatus for restarting a computing system

• US 4,648,031 A
• Filed: 06/21/1982
• Issued: 03/03/1987
• Est. Priority Date: 06/21/1982
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Jenner, Earle H.
• Primary Examiner(s): Heckler, Thomas M.
• Application Number:
  US05/390,163
• Time in Patent Office: 1,716 Days
• Field of Search: 364/200 MS File, 364/900 MS File
• US Class Current: 714/10
• CPC Class Codes: G06F 11/1441 Resetting or repowering
• CPC Class Codes: G06F 16/00 Information retrieval; Database structures therefor; File system structures therefor
• CPC Class Codes: G06F 9/463 Program control block organisation

Abstract

A programming method and structure for operating a computing system to restart a total subsystem or a subset of that subsystem to an operable state following a total interruptions (system failure or termination, either normal or abnormal). The subsystem isolates inoperable resources while permitting the others to resume by independently maintaining in a first structure recording the completion state of a resource manager'"'"'s recovery responsibility with respect to each interrupted work unit and in a second structure the operational states and recovery log interest scopes of each resource manager. The completion state can be influenced by the starting or not of a resource manager, and if restarted, the presence or absence of a resource subset required to accomplish the work unit recovery.

Claims

1. Apparatus for restarting a computing system following interruption, including a plurality of resource managers that manage resource collections such as data bases, non-volatile storage means containing a recovery log on which are recorded checkpointed states and images of changes resulting from the execution of work unit instructions, a subset of said resource managers having responsibilities in the execution of work unit instructions and the recovery of work units during restart, said apparatus comprising:

first data structure means for recording for an interrupted work unit the location of a respective section of the recovery log which contains information of its activities and the recovery responsibility of each resource manager to the work unit;

second data structure means for recording for each resource manager its operational state and the location of a respective different section of the recovery log which contains information that particular resource manager needs to perform its recovery responsibility; and

recovery means which receives and is responsive to information recorded in said first and second data structure means for restarting selected resource managers following interruption and deferring the restarting of other resource managers.

2. The apparatus of claim 1 wherein said recovery means further includes means for periodically recording on the non-volatile storage means a copy of the first data structure means when, at the conclusion of a restart operation, the recovery responsibilities of at least one resource manager have not been completed.

3. The apparatus of claim 2 wherein said recovery means further includes means for periodically recording on the non-volatile storage means a copy of said second data structure means and of changes thereto.

4. The apparatus of claim 3 wherein said recovery means further includes means for reading log records during a recovery operation to build from information therein said first data structure means.

5. The apparatus of claim 4, wherein said recovery means further includes means for reestablishing during a first restart phase the state of the resource collection managed by each resource manager being restarted, beginning with the most recent checkpointed state and then updated to the point of interruption using records of changes in the log.

6. The apparatus of claim 5, wherein said recovery means further includes means for extending, during a second restart phase for each resource manager being restarted, the state of the resource collections it manages to include those changes such resource manager is responsible for, by reallocating resource collections exclusively to work units interrupted while in a state wherein it was unknown whether the work units had been committed or not, and by writing into said non-volatile storage the most recent states of the resource collections which had been committed before interruption.

7. The apparatus of claim 6, wherein said recovery means further includes means for restoring during a third restart phase for each resource manager being restarted the resource collections altered by interrupted work units to return to the state to which it had most recently been committed.

8. A method for restarting a computing system following interruption, including a plurality of resource managers that manage resource collections such as data bases, a recovery log in non-volatile storage on which are recorded checkpointed states and images of changes resulting from the execution of work unit instructions, a subset of said resource managers having responsibilities in the execution of work unit instructions and the recovery of work units during restart, said method comprising the steps of:

recording in a first data structure means for an interrupted work unit the location of a respective section of the recovery log which contains information of its activities and the recovery responsibility of each resource manager to the interrupted work unit;

recording in a second data structure means for each resource manager the operational state of such resource manager and the location of a respective section of the recovery log which contains information such resource manager needs to perform its recovery responsibility; and

responsive to information recorded in said first and second data structure means, restarting selected resource managers following interruption while deferring the restarting of other resource managers.

9. The method of claim 8 further including the steps of periodically recording on non-volatile storage a copy of the first data structure means when, at the conclusion of a restart operation, the recovery responsibilities of at least one resource manager have not been completed.

10. The method of claim 9 including the further steps of periodically recording on non-volatile storage a copy of said second data structure means and of changes thereto.

11. The method of claim 10 further including the step of reading log records during a recovery operation to build from information therein said first data structure means.

12. The method of claim 11, further including the step of reestablishing during a first restart phase the state of the resource collection managed by each resource manager being restarted, beginning with the most recent checkpointed state and then updated to the point of interruption using records of changes in the log.

13. The method of claim 12, further including the step of updating, during a second restart phase for each resource manager being restarted, the state of the resource collections it manages to include those changes such resource manager is responsible for, by reallocating resource collections exclusively to work units interrupted while in a state wherein it was unknown whether the work units had been committed or not, and by writing into said non-volatile storage the most recent state of the resource collections which had been committed before interruption.

14. The rethod of claim 13, further including the step of restoring during a third restart phase for each resource manager being restarted the resource collections altered by interrupted work units to the state which had most recently been comritted.

15. A method for facilitating the restarting of a computing system that is interrupted from time to time, the system having a plurality of resource managers that manage resource collections such as data bases, a recovery log in non-volatile storage on which are recorded checkpointed states and images of changes resulting from the execution of work unit instructions, a subset of said resource managers having responsibilities in the execution of work unit instructions and the recovery of work units after interruption, said method comprising the steps of:

(a) recording for each interrupted work unit the location of a respective section of the recovery log which contains information of its activities and the recovery responsibility of each resource manager to the work unit;

(b) recording for each resource manager of said subset;

(1) its operational state and (2) the location of a respective different section of the recovery log which contains the information that particular resource manager needs to perform its recovery responsibility; and

(c) following each interrupt, restarting at least one resource manager in response to the information recorded in steps (a) and (b).

16. The method of claim 15 wherein the restarting step (c) comprises the further steps of:

reestablishing the state of the resource collections managed by each resource manager being restarted, beginning with the most recent checkpointed state and then updated to the point of interruption using records of changes in the log;

extending, for each resource manager being restarted, the state of the resource collections such resource manager manages to include those changes it is responsible for, by reallocating resource collections exclusively to work units interrupted while in states wherein it is unknown whether the work units have been committed or not, and by writing into said non-volatile storage the most recent states of the resource collections which had been committed before interruption; and

restoring, for each resource manager being restarted, the resource collections altered by interrupted work units to return to the state to which it had most recently been committed.