Autonomous cluster organization, collision detection, and resolutions

• US 7,788,522 B1
• Filed: 05/31/2007
• Issued: 08/31/2010
• Est. Priority Date: 05/31/2007
• Status: Active Grant

• Current Assignee: Oracle America, Inc. (Oracle Corporation)
• Original Assignee: Oracle America, Inc. (Oracle Corporation)
• Inventors: White, Lawrence, Ganapathy, Shreedhar, Abdelaziz, Mohamed M., Traversat, Bernard, Kumar, Abhijit
• Primary Examiner(s): McCarthy; Christopher S
• Application Number:
  US11/756,480
• Time in Patent Office: 1,188 Days
• Field of Search: None
• US Class Current: 714/4.1
• CPC Class Codes: G06F 11/1425 by reconfiguration of node membership
• CPC Class Codes: H04L 41/12 Discovery or management of network topologies
• CPC Class Codes: H04L 67/104 Peer-to-peer [P2P] networks
• CPC Class Codes: H04L 67/1051 Group master selection mechanisms

Abstract

A method, apparatus, and system are directed toward managing cluster membership through a discovery process that enables selection of a master for the cluster. If a node on a network is the first to initiate the discovery process, it assumes the master role. Where two or more nodes enter the discovery process at the same time, or where a cluster is disjointed, views of the potential members to the cluster are shared among the nodes. Each node coalesces and sorts the views. The master may then be determined from the sorted views. In one embodiment, the master is that node in the results having a highest unique identifier. The identified master sends an assertion of its role to the other nodes. If no assertion is received, or multiple assertions are received, updated views are again shared and sorted. In one embodiment, the nodes may be within a peer-to-peer network.

Claims

1. A network computing device for managing a cluster network, comprising:

a processor;

a memory for use in receiving and managing data and instructions that, when executed by the processor performs a method, the method comprising;

examining a first view of the cluster network to identify a candidate master node for the cluster network based on a sorted order of cluster member identifiers within the first view;

when the network computing device is identified as the candidate master node, sending an announcement to other cluster members of the cluster network indicating that the network computing device is a first asserted master node; and

upon detecting a master collision, wherein the master collision indicates a second asserted master node;

obtaining a first updated cluster view of the cluster network from each of the cluster members of the cluster network,

coalescing the received updated cluster views to obtain a second view of the cluster network,

sorting the cluster member identifiers within the second view of the cluster network to obtain a sorted second view of the cluster network, and

identifying a master node based on the sorted second view of the cluster network.

2. The network computing device of claim 1, wherein the method further comprises:

sending a master node query message over the cluster network to determine if the master node already is identified; and

when a master response is received within a defined time period,

receiving at least one authoritative cluster view from the master node for the cluster network, wherein the authoritative cluster view comprises information about the networking computing device, and identifies the master node.

3. The network computing device of claim 1, wherein identifying the candidate master node is based on a first cluster member identifier within the sorted second view of the cluster network.

4. The network computing device of claim 1, wherein sorting the cluster member identifiers comprises one selected from a group consisting of sorting from lowest to highest identifiers and sorting from highest to lowest identifiers.

5. The network computing device of claim 1, wherein the method further comprises:

receiving a health message from a first cluster member of the cluster network;

determining that the health message indicates that the first cluster member within the cluster network is failed,

determining that the network computing device is not the master node, and

obtaining an updated authoritative cluster view from the master node that excludes information about the first cluster member.

6. The network computing device of claim 1, wherein the processor configured to execute the instructions, and performed actions, further including:

when the master collision is undetected, sending an announcement that the network computing device is the master node for the cluster network; and

assuming a cluster management role for the cluster network, including at least sending one authoritative cluster view to other cluster members of the cluster network.

7. The network computing device of claim 1, wherein the method further comprises:

receiving a health message from a first cluster member of the cluster network;

determining that the health message indicates that the master node is failed,

receiving a second updated cluster view from each of the other cluster members of the cluster network;

coalescing and sorting the received second updated cluster views to generate a third cluster view; and

employing the third cluster view to identify a new candidate master node for the cluster network.

8. The network computing device of claim 1, wherein the first updated cluster views are received using a cluster view advertisement.

9. The network computing device of claim 1, wherein the cluster member identifiers are generated using a secure hash algorithm based on the string based name for a cluster member.

10. The network computing device of claim 1, wherein the first cluster view comprises at least an identifier for the cluster network, and at least one cluster member identifier.

11. The network computing device of claim 1, wherein the method further comprises:

monitoring for health messages from other members within the cluster network;

when a health message is not received within a defined time period for a cluster member, and the network computing device is the master node for the cluster network, sending a message to the other cluster members of the cluster network indicating that the cluster member is in doubt; and

when a failure verification interval is exceeded for the cluster member, sending an authoritative cluster view to the other cluster members of the cluster network indicating the cluster member in doubt has failed.

12. A computer readable medium comprising computer readable code for managing a network cluster that, when executed by a processor is configured to employ a method, the method comprising:

examining a first view of the cluster network to identify a candidate master node for the cluster network based on a sorted order of cluster member identifiers within the first view;

when the network computing device is identified as the candidate master node, sending an announcement to other cluster members of the cluster network indicating that the network computing device is a first asserted master node; and

upon detecting a master collision, wherein the master collision indicates a second asserted master node,

obtaining a first updated cluster view of the cluster network from each of the cluster members of the cluster network,

coalescing the received updated cluster views to obtain a second view of the cluster network,

sorting the cluster member identifiers within the second view of the cluster network to obtain a sorted second view of the cluster network, and

identifying a master node based on the sorted second view of the cluster network.

13. The computer readable storage medium of claim 12, the method further comprising:

receiving a health message from a first cluster member of the cluster network;

determining that the health message indicates that the master node is failed,

receiving a second updated cluster view from each of the other cluster members of the cluster network;

coalescing and sorting the received updated cluster views to generate a third cluster view; and

employing the third cluster view to identify a new candidate master node for the cluster network.

14. The computer readable storage medium of claim 12, wherein the first cluster view comprises at least an identifier for the cluster network, and at least one cluster member identifier.

15. The computer readable storage medium of claim 12, wherein the first updated cluster views are received using a cluster view advertisement.

16. The computer readable storage medium of claim 12, wherein sorting the cluster member identifiers comprises one selected from a group consisting of sorting from lowest to highest identifiers, and sorting from highest to lowest identifiers.

17. A system for use in managing a cluster, comprising:

a first node that is configured to perform actions, including;

examining a first view of the cluster network to identify a candidate master node for the cluster network based on a sorted order of cluster member identifiers within the first view;

when the first node is identified as the candidate master node, sending an announcement to other cluster members of the cluster network indicating that the first cluster member is a first asserted master node; and

upon detecting a master collision, wherein the master collision indicates that another cluster member is a second asserted master node,

receiving an updated cluster view of the cluster network from each of the cluster members,

coalescing the received updated cluster views with the first node'"'"'s cluster view into a second cluster view,

sorting the identifiers within the second cluster view to obtain a sorted second view of the cluster network, and

identifying a first master node based on the sorted second view of the cluster network; and

a second node that is configured to perform actions, including;

examining a third view of the cluster network to identify the candidate master node for the cluster network based on a sorted order of cluster member identifiers within the third view; and

when the second node is identified as the candidate master node, and the first node is the first asserted master node, then;

sending a master collision message to the first node;

sending the third view of the cluster network to the first node;

receiving the second view from the first node;

coalescing the second view and the third view to obtain a fourth view,

sorting the identifiers within the fourth view to obtain a sorted fourth view of the cluster network, and

identifying the master node based on the sorted fourth view of the cluster network.

18. The system of claim 17, wherein the cluster network employs a peer-to-peer architecture.

19. The system of claim 17, wherein the first node and the second node monitor for health messages from members within the cluster network, and if a cluster member node is determined to be failed, receiving from the identified master node an authoritative cluster view of the cluster network that excludes the failed cluster member'"'"'s identifier.

20. The system of claim 17, wherein one selected from a group consisting of the first node and the second node, during initialization into the cluster network, transmits a master node query to determine if an existing master node is already identified for the cluster network, wherein the identification of the existing master node is determined, at least in part, based on receiving a master response to the query within a defined time period.