×

Managing checkpoint queues in a multiple node system

  • US 7,065,540 B2
  • Filed: 03/04/2002
  • Issued: 06/20/2006
  • Est. Priority Date: 11/24/1998
  • Status: Expired due to Term
First Claim
Patent Images

1. A computer executable method of managing information about where to begin recovery after a failure of one or more nodes of a multiple-node system, the method comprising the steps of:

  • in a particular node of the multiple-node system, maintaining botha single-failure queue indicating where within a recovery log to begin recovery after a failure of said node; and

    a multiple-failure queue indicating where within said recovery log to begin recovery after a failure of said node and one or more other nodes in said multiple-node system;

    in response to a dirty data item being written to persistent storage, removing an entry for said data item from both said single-failure queue and said multiple-failure queue; and

    in response to a dirty data item being sent to another node of said multiple-node system without first being written to persistent storage, removing an entry for said data item from said single-failure queue without removing the entry for said data item from said multiple-failure queue.

View all claims
  • 2 Assignments
Timeline View
Assignment View
    ×
    ×