Graceful failover using augmented stubs

• US 7,434,087 B1
• Filed: 05/21/2004
• Issued: 10/07/2008
• Est. Priority Date: 05/21/2004
• Status: Active Grant

• Current Assignee: Oracle America, Inc. (Oracle Corporation)
• Original Assignee: Sun Microsystems Incorporated (Oracle Corporation)
• Inventors: Singh, Servesh Pratap
• Primary Examiner(s): Lohn; Joshua A
• Application Number:
  US10/851,844
• Time in Patent Office: 1,600 Days
• Field of Search: 714/4
• US Class Current: 714/4.11
• CPC Class Codes: G06F 11/1425 by reconfiguration of node membership
• CPC Class Codes: G06F 11/1443 Transmit or communication errors
• CPC Class Codes: G06F 11/2025 using centralised failover control functionality
• CPC Class Codes: G06F 11/2041 with more than one idle spare processing component
• CPC Class Codes: G06F 11/2046 where the redundant components share persistent storage G06F11/2043 takes precedence

Abstract

Various embodiments of a system and method for failing over one cluster node to another are disclosed. The system may include a client networked to a cluster of computer nodes, connected to a persistent data store such as a database resident on a backend system. Each node of the cluster may be configured to run a distributed application component. The client may be configured to communicate with an instance of the application running on a first node. The client may include an augmented stub capable of accessing the instance of the application on the first node and capable of detecting a failure of the first node. In response to detecting a failure in the first node, the augmented stub may be capable of failing over to a second node and accessing the instance of the application on the second node, transparently to the client.

Claims

1. A system, comprising:

a cluster, comprising a plurality of nodes configured to execute a distributed application component;

a client configured to generate a lookup request for the application component;

a naming service configured to respond to the lookup request, wherein the naming service is configured to, in response to the lookup request;

determine whether the application component is executing on the cluster;

return an augmented stub if the application component is executing on the cluster; and

return a normal stub if the application component is not executing on the cluster;

wherein the client is configured to access the application component using the augmented stub; and

the augmented stub is configured to;

interface to an instance of the application component on one of the nodes;

in response to a failure of the one of the nodes, failover to another one of the nodes, transparent to the client; and

interface to another instance of the application component on the another one of the nodes.

2. The system, as recited in claim 1, wherein the augmented stub comprises a stub to communicate with the one of the nodes and an indication of other ones of the nodes to which the application component can be failed over, wherein to failover, the augmented stub is configured to:

select the another one of the nodes based on the indication;

obtain a new stub corresponding to the another one of the nodes;

replace the stub with the new stub; and

retry and/or resume communication using the new stub.

3. The system, as recited in claim 1, wherein the augmented stub comprises a plurality of stubs, wherein one of the stubs is configured to communicate with the one of the nodes and each of the other stubs is configured to communicate with one of the other nodes of the plurality of nodes to which the application component can be failed over, wherein to failover, the augmented stub is configured to:

select a new stub from the other stubs; and

retry and/or resume communication using the new stub.

4. The system, as recited in claim 1, wherein each cluster node comprises node cluster membership data comprising a node unique identifier indicative of cluster membership during a period of time, and wherein each cluster node is configured to detect a change in cluster membership and, in response to said detecting, to update the cluster membership data including the unique identifier.

5. The system, as recited in claim 4, wherein the augmented stub comprises client cluster membership data comprising a client unique identifier indicative of cluster membership during a period of time and is configured to:

send the client identifier to one of the plurality of nodes;

receive node cluster membership data; and

update the client cluster membership data based on the node cluster membership data.

6. A method, comprising:

a client generating a lookup request to a naming service for an application component;

the naming service responding to the lookup request, wherein the naming service responding to the lookup request comprises;

determining whether the application component is executing on a cluster;

returning an augmented stub if the application component is executing on a cluster; and

returning a normal stub if the application component is not executing on a cluster;

the client accessing the application component using the augmented stub;

the augmented stub interfacing to a first instance of the application component on a first node of a cluster;

in response to a failure of the first node, the augmented stub failing over to a second node of the cluster transparent to the client; and

the augmented stub interfacing to a second instance of the application component on the second node.

7. The method, as recited in claim 6, wherein the augmented stub comprises a stub to communicate with the first node and data on other nodes of the cluster to which the application component can be failed over, wherein failing over comprises:

selecting the second node based on the data;

obtaining a new stub corresponding to the selected node;

replacing the stub with the new stub; and

retrying and/or resuming communication using the new stub.

8. The method, as recited in claim 6, wherein the augmented stub comprises a plurality of stubs, wherein one of the stubs is configured to communicate with the first node and each of the other stubs is configured to communicate with one of the other nodes of the cluster to which the application component can be failed over, wherein failing over comprises:

selecting a new stub from the other stubs; and

retrying and/or resuming communication using the new stub.

9. The method, as recited in claim 6, wherein each cluster node comprises node cluster membership data comprising a node unique identifier indicative of cluster membership during a period of time, further comprising each cluster node detecting a change in cluster membership and, in response to said detecting, updating the cluster membership data including the unique identifier.

10. The method, as recited in claim 9, further comprising:

the augmented stub sending to one of the plurality of nodes a unique identifier representing cluster membership information last received by the augmented stub;

comparing the unique identifier from the augmented stub to the node'"'"'s unique identifier indicative of current cluster membership; and

updating cluster membership to the augmented stub if said comparing indicates a change in cluster membership.

11. A method comprising:

receiving a lookup request for an application component from a client;

determining whether the application component is executing on a cluster;

in response to the lookup request;

returning an augmented stub for interfacing to the application component if the application component is executing on a cluster, wherein the augmented stub is configured to provide fail-over for the application component to another node in the cluster transparently to the client; and

returning a normal stub for interfacing to the application component if the application component is not executing on a cluster.

12. The method, as recited in claim 11, wherein said receiving, determining, and returning is performed by a naming service.

13. A computer-accessible medium comprising program instructions, wherein the program instructions are computer-executable to implement:

a client configured to generate a lookup request for an application component;

a naming service configured to respond to the lookup request, wherein the naming service is configured to, in response to the lookup request;

determine whether the application component is executing on a cluster;

return an augmented stub if the application component is executing on a cluster; and

return a normal stub if the application component is not executing on a cluster

wherein the client is configured to access the application component using the augmented stub;

the augmented stub configured to;

interface to an instance of the application component on a first node of a cluster;

in response to a failure of the first node, failover to a second node of the cluster, transparent to the client; and

interface to another instance of the application component on the second node.

14. The computer-accessible medium, as recited in claim 13, wherein the augmented stub comprises a stub to communicate with the first node and data on other nodes of the cluster to which the application component can be failed over, wherein to failover, the augmented stub is configured to:

select the select the second node based on the data;

obtain a new stub corresponding to the second node;

replace the stub with the new stub; and

retry and/or resume communication using the new stub.

15. The computer-accessible medium, as recited in claim 13, wherein the augmented stub comprises a plurality of stubs, wherein one of the stubs is configured to communicate with the node and each of the other stubs is configured to communicate with one of the other nodes of the plurality of nodes to which the application component can be failed over, wherein to failover, the augmented stub is configured to:

select a new stub from the other stubs; and

retry and/or resume communication using the new stub.

16. The computer-accessible medium, as recited in claim 13, wherein each cluster node comprises node cluster membership data comprising a node unique identifier indicative of cluster membership during a period of time, and wherein each cluster node is configured to detect a change in cluster membership and, in response to said detecting, to update the cluster membership data including the unique identifier.

17. The computer-accessible medium, as recited in claim 16, wherein the augmented stub is configured to:

send to one of the plurality of nodes a unique identifier representing cluster membership information last received by the augmented stub; and

receive an update of cluster membership if the unique identifier from the augmented stub differs from the node'"'"'s unique identifier indicative of current cluster membership.