Synchronizing cluster time

• US 20080183894A1
• Filed: 01/25/2007
• Published: 07/31/2008
• Est. Priority Date: 01/25/2007
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Srivastava, Alok, Rai, Vikram
• Granted Patent: US 7,814,360 B2
• Time in Patent Office: Days
• US Class Current: 709/248
• CPC Class Codes: H04J 3/0641 Change of the master or reference, e.g. take-over or failure of the master
• CPC Class Codes: H04L 69/28 Timers or timing mechanisms used in protocols

Abstract

Synchronizing cluster time is described. Times of one or more nodes in the cluster to a master node of the cluster are synchronized. It is discovered that a node of a cluster has a clock time that is ahead of a master clock time of the master node. In response to discovering the node with the clock time ahead of the master clock time, the master node is replaced with the node ahead thereof in time, which thus becomes a new master node. The clock times of the cluster nodes are synchronized with the clock time of the new master node.

Claims

12-1. The method as recited in claim 1 wherein, upon initial formation of the cluster, a first node to join the cluster becomes the master and, upon a subsequent node joining the cluster, the subsequent node becomes a slave.

14. A method for synchronizing time in a computer cluster;

comprising the steps of;

monitoring the clock times of the cluster nodes in relation to that of a first master node;

discovering that a node of a cluster has a clock time that is ahead of a master clock time of the first master node;

ascertaining that the node that is ahead of the first master clock in time is ahead in time to a degree that satisfies one or more criteria;

in response to ascertaining that the node is ahead, atomically transferring a master status from the first master node to the node ahead thereof in time wherein the node ahead in time becomes a new master node of the cluster and the clock time of the new master node becomes a new master clock time of the cluster;

synchronizing the clock times of other nodes in the cluster with the new master clock time; and

iteratively repeating the transferring step based on the discovering step.

2. The method as recited in claim 1 wherein the replacing step is performed atomically between the original master node and the node ahead thereof in time.

15. The method as recited in claim 14 wherein the ascertaining step comprises the steps of:

logging the time kept with each clock of nodes of the cluster;

computing a mean cluster time based on the logging step; and

determining that the clock time of the node ahead of the pre-existing master clock is within a range of the mean cluster time.

17. A computer readable medium having instructions encoded therewith which, when executed with one or more processors of a computer system, cause the processors to execute the method recited in claim 2.

26. A computer readable medium having instructions encoded therewith which, when executed with one or more processors of a computer system, cause the processors to execute the method recited in claim 11.

29. A computer readable medium having instructions encoded therewith which, when executed with one or more processors of a computer system, cause the processors to execute the method recited in claim 14.

26-2. A computer readable medium having instructions encoded therewith which, when executed with one or more processors of a computer system, cause the processors to execute the method recited in claim 15.