Methods and devices for recovering from initialization failures

• US 20060294507A1
• Filed: 06/28/2005
• Published: 12/28/2006
• Est. Priority Date: 06/28/2005
• Status: Active Grant

• Current Assignee: Alcatel-Lucent USA, Inc. (Nokia Corporation)
• Original Assignee: Alcatel-Lucent USA, Inc. (Nokia Corporation)
• Inventors: Buskens, Richard Wayne, Gonzalez, Oscar, Ren, Yansong
• Granted Patent: US 7,707,451 B2
• Time in Patent Office: Days
• US Class Current: 717/133
• CPC Class Codes: G06F 11/1417 Boot up procedures
• CPC Class Codes: G06F 9/5066 Algorithms for mapping a plurality of inter-dependent sub-tasks onto a plurality of physical CPUs mappping at compile time, see G06F8/451

Abstract

The time it takes to recover from a system initialization failure may be reduced by determining whether to enable a recovery process immediately, or defer such a process. Sometimes it is desirable to defer a recovery process until certain interdependencies between system components are satisfied.

Claims

1. A method for recovering from an initialization failure comprising:

generating a recovery graph to identify a set of initialization tasks to execute or re-execute in the event of an initialization failure.

2. The method as in claim 1 wherein the recovery graph comprises a recomputed interdependency graph.

3. The method as in claim 1 further comprising generating the recovery graph based on types of dependencies present in an interdependency graph and in execution states of initialization tasks.

6. The method as in claim 1 further comprising deferring recovery until a later stage of an initialization process to realize a faster initialization time.

7. The method as in claim 6 further comprising deferring recovery until one or more tasks have completed their utilization of one or more other tasks they depend upon.

4. A method for generating a recovery graph comprising:

adding initialization tasks associated with failed components in a recovery graph;

adding already-executed tasks into the recovery graph;

adding tasks already in progress or not yet executed to the graph; and

adding dependencies copied from an original interdependency graph into the recovery graph.

5. A method for generating a recovery graph comprising adding dependencies into a recovery graph, where the dependencies are selected from at least the group consisting of operational, sequential, fail-over, non-essential, unmanaged, and fault escalating.

8. A method for determining whether to immediately initiate recovery from an initialization failure or defer such recovery comprising:

identifying execution times for each initialization task;

estimating remaining initialization times associated with an immediate recovery scenario and a deferred recovery scenario; and

selecting the scenario that provides the smaller, overall initialization time.

9. A method for estimating an expected, remaining initialization time comprising:

generating a first recovery graph;

calculating a remaining initialization time for each component within the graph;

computing a first maximum initialization time as an immediate recovery time;

identifying tasks currently using services provided by non-faulty components;

determining a remaining time for each identified task to complete execution;

generating a second recovery graph containing the identified tasks;

calculating a remaining initialization time for each component in the second graph; and

computing a second maximum initialization time as a deferred recovery time.

10. A method for recovering from an initialization failure comprising:

generating an interdependency graph comprising initialization tasks;

determining a recovery scheme that minimizes an overall initialization time; and

triggering initialization activity associated with initialization tasks.

11. The method as in claim 10 further comprising:

immediately recovering from an initialization failure by terminating processes associated with faulty components, computing a recovery graph and continuing initialization using the recovery graph.

12. The method as in claim 10 further comprising:

deferring recovery from an initialization failure by placing certain tasks on hold, and determining whether to continue to defer recovery after each task, that has not been placed on hold, has been executed.

13. A device for recovering from an initialization failure operable to:

generate a recovery graph to identify a set of initialization tasks to execute or re-execute in the event of an initialization failure.

14. The device as in claim 13 further operable to defer recovery until a later stage of an initialization process to realize a faster initialization time.

15. The device as in claim 13 further operable to defer recovery until one or more tasks have completed their utilization of one or more other tasks they depend upon.

16. The device as in claim 13 further operable to:

identify execution times for each initialization task;

estimate remaining initialization times associated with an immediate recovery scenario and a deferred recovery scenario; and

select the scenario that provides the smaller, overall initialization time.

17. The device as in claim 13 further operable to:

generate an interdependency graph comprising initialization tasks;

determine a recovery scheme that minimizes an overall initialization time; and

trigger initialization activity associated with initialization tasks.

18. The device as in claim 17 further operable to:

immediately recover from an initialization failure by terminating processes associated with faulty components, computing the recovery graph and continuing initialization using the recovery graph.

19. The device as in claim 17 further operable to:

defer recovery from an initialization failure by placing certain tasks on hold, and determine whether to continue to defer recovery after each task, that has not been placed on hold, has been executed.