Maintaining state information in a multi-component, event-driven state machine

Patent metadata

Field	Value
Publication number	US-9588780-B2
Application number	US-201514831141-A
Country	US
Kind code	B2
Filing date	Aug 20, 2015
Priority date	Dec 9, 2014
Publication date	Mar 7, 2017
Grant date	Mar 7, 2017

Abstract

A method, apparatus and computer program product that allows for maintaining correct states of all sub-components in a state machine, even as sub-components leave the state machine and later rejoin in some previous state. Preferably, this is achieved without requiring the system to remember the states of all sub-components or a log of every event that was fed into the state machine. Thus, the technique does not require any knowledge of the previous state of the sub-components nor the need to preserve a complete log of events that were fed into the state machine. The state machine may be used to enhance the operation of a technological process, such as a workload management environment.

First claim

The invention claimed is:

1. A method for managing a state machine comprising a main component, and one or more sub-components, wherein the main component maintains, for each sub-component, a log of events associated with transitions in states of the main component and the one or more sub-components, comprising:
for each sub-component, and in response to occurrence of an event that causes the sub-component to transition state, determining whether the event supersedes or reverses a prior event previously logged in association with a state transition of the sub-component;
upon determining that the event supersedes or reverses the prior event previously logged in association with a state transition, pruning the prior event from the log to generate a pruned log;
upon joinder or re-joinder to the state machine of a given sub-component, and prior to the main component initiating a replay operation for the given sub-component, forcing the given sub-component into an initial state; and
thereafter using the pruned log to transition the given sub-component from its initial state to a time-consistent state associated with the main component and each of the one or more sub-components, wherein the pruned log is used to transition the given sub-component from its initial state to the time-consistent state during the replay operation, the replay operation replaying a list of events recorded by the main component while the given sub-component is not connected to the state machine; and
wherein the state machine does not need to have any knowledge of a current state of the given sub-component at a time the given sub-component connects to the state machine.

2. The method as described in claim 1 wherein the given sub-component is placed into the initial state using one and only one event.

3. The method as described in claim 1 further including recording the triggering event in the log.

4. The method as described in claim 1 further including saving the log and the pruned log in association with a backup main component.

5. The method as described in claim 4 further including transitioning the backup main component to function as the main component upon a failure of the main component.

6. The method as described in claim 1 wherein the state machine manages operational states for a workload management environment.

Assignees

• IBM

Inventors

• Fitzpatrick Michael Gerard
• Arrowood Andrew Hilliard
• Mcafee Gary Owen
• Huang Sue L

Classifications

• G06F9/542

  Event management; Broadcasting; Multicasting; Notifications · CPC title

• G06F9/4498 Primary

  Finite state machines · CPC title

• G06F9/5083 Primary

  Techniques for rebalancing the load in a distributed system · CPC title

• G06F9/444 Primary

  Physics · mapped topic