Fault tolerant listener registration in the presence of node crashes in a data grid

Patent metadata

Field	Value
Publication number	US-2016124817-A1
Application number	US-201514609329-A
Country	US
Kind code	A1
Filing date	Jan 29, 2015
Priority date	Oct 31, 2014
Publication date	May 5, 2016
Grant date	—

Abstract

A processing device to perform operations comprising receive, from a listener of a second node in a data grid system, a filter defined by search criteria of a search query. The operations can include determine, at the first node, that a third node in the data grid has crashed. The operations can further include iterate over backup data of the third node that is stored at a memory of the first node to determine the backup data that matches the filter. The operations can further include communicate, to the listener, the backup data that matches the filter.

First claim

What is claimed is:

1 . A non-transitory computer readable storage medium having instructions that, when executed by a processing device of a first node, cause the processing device to perform operations comprising:
receive, from a listener of a second node in a data grid system, a filter defined by search criteria of a search query; determine, at the first node, that a third node in the data grid has crashed; iterate, by the processing device of the first node, over backup data of the third node that is stored at a memory of the first node to determine the backup data that matches the filter; and communicate, to the listener, the backup data that matches the filter.

2 . The non-transitory computer readable storage medium of claim 1 , the operations further comprising communicate, to the listener, a duplicate data notification, wherein the duplicate data notification indicates that the communicated backup data may comprise, at least in part, data duplicative to filtered data communicated to the listener from the third node.

3 . The non-transitory computer readable storage medium of claim 1 , the operations further comprising communicate, to the listener, data from a memory coupled to the processing device that matches the filter.

4 . The non-transitory computer readable storage medium of claim 1 , wherein the receive, determine, iterate, and communicate operations occur during an iteration period of the search query.

5 . The non-transitory computer readable storage medium of claim 1 , wherein the first node, the second node, and the third node are part of a node cluster in the data grid system.

6 . The non-transitory computer readable storage medium of claim 1 , the operations further comprising remove, at the first node, filters received from the third node when the processing device determines that the third node has crashed.

7 . The non-transitory computer readable storage medium of claim 1 , wherein the operation of determining that the third node has crashed further comprises:
receive, from the third node, heartbeat messages at periodic intervals, wherein each heartbeat message indicates that the third node is active; determine that a heartbeat message has not been received at the first node within a threshold period of time; communicate, to a fourth node in the data grid, a check node request, wherein the check node request requests that the fourth node verify that the third node has crashed; receive, from the fourth node, a crash notification when the third node has not responded to the check node request within a threshold period of time; and determine that the third node has crashed in view of the crash notification.

8 . The non-transitory computer readable storage medium of claim 1 , the operations further comprising rebuild a queue of modifications for the third node when the third node has crashed.

9 . The non-transitory computer readable storage medium of claim 8 , the operations further comprising communicating the rebuilt queue to the listener.

10 . A system comprising:
a memory to store data; and a processor coupled to the memory, the processor to perform operations comprising:
receiving, from a listener of a second node in a data grid system, a filter defined by search criteria of a search query;
determining, at the first node, that a third node in the data grid has crashed;
rebuilding, at the first node, a queue of changes to data in a memory of the third node when the third node has crashed; and
communicating, to the listener, the rebuilt queue of changes to the data of the third node.

11 . The system of claim 10 , the operations further comprising communicate, to the listener, data from a memory coupled to the processing device that matches the filter.

12 . The system of claim 10 , the operations further comprising:
iterate over, at the first node, backup data of the third node that is stored at the memory to determine the backup data that matches the filter; and communicate, to the listener, the backup data that matches the filter.

13 . The system of claim 10 , wherein the rebuilt queue of changes to the data of the third node occurs during an iteration period of the search query.

14 . The system of claim 10 , the operations further comprising record, into a queue, modifications to the memory of the first node during an iteration period.

15 . The system of claim 14 , the operations further comprising communicating the queue of the first node to the listener.

16 . The system of claim 10 , wherein the operation of determining that the third node has crashed further comprises:
receiving, from the third node, heartbeat messages at periodic intervals, wherein each heartbeat message indicates that the third node is active; determining that a heartbeat message has not been received at the first node within a threshold period of time; communicating, to a fourth node in the data grid, a check node request, wherein the check node request requests that the fourth node verify that the third node has crashed; receiving, from the fourth node, a crash notification when the third node has not responded to the check node request within a threshold period of time; and determining that the third node has crashed in view of the crash notification.

17 . A method comprising:
communicating, from a listener of a first node in a data grid, a data filter to a second node and a third node in a data grid system; receiving, at the listener, data from a memory of the second node that matches the data filter; receiving, at the listener, backup data from a memory of the third node that matches the data filter; receiving, at the listener, a duplicate data notification from the third node indicating that the backup data may contain data duplicative of the data received by the listener from the second node; and combining, by a processing device executing the listener, the data from the memory of the second node and the backup data from the memory of the third node, wherein the combining comprises adding, to the data from the memory of the second node, a portion of the backup data from the memory of the third node that does not contain the duplicative data.

18 . The method of claim 17 , wherein:
the backup data is data of the second node stored at the third node, and the backup data is communicated by the third node when the second node has crashed.

19 . The method of claim 17 , further comprising removing a duplicative data point of the duplicative data using an idempotent function.

20 . The method of claim 17 , further comprising receiving, at the listener when the second node has crashed, a rebuilt queue of changes to data of the second node.

Assignees

• Red Hat Inc

Inventors

• Markus Mircea
• Burns William Rosenquist

Classifications

• G06F11/2043

  where the redundant components share a common memory address space · CPC title

• G06F11/2025

  using centralised failover control functionality · CPC title

• G06F16/2358

  Change logging, detection, and notification (replication G06F16/27) · CPC title

• G06F16/2255

  Hash tables · CPC title

• G06F16/2291

  User-Defined Types; Storage management thereof · CPC title