US 11,734,133 B2
Cluster system and fail-over control method of cluster system
Keita Sugihara, Tokyo (JP); Keisuke Matsumoto, Tokyo (JP); and Yuki Kuroda, Tokyo (JP)
Assigned to HITACHI, LTD., Tokyo (JP)
Filed by Hitachi, Ltd., Tokyo (JP)
Filed on Aug. 17, 2021, as Appl. No. 17/404,469.
Claims priority of application No. 2020-180031 (JP), filed on Oct. 27, 2020.
Prior Publication US 2022/0129357 A1, Apr. 28, 2022
Int. Cl. G06F 11/20 (2006.01); G06F 11/30 (2006.01); G06F 3/06 (2006.01); G06F 9/455 (2018.01)
CPC G06F 11/2023 (2013.01) [G06F 3/067 (2013.01); G06F 3/0619 (2013.01); G06F 3/0652 (2013.01); G06F 3/0665 (2013.01); G06F 9/45558 (2013.01); G06F 11/3006 (2013.01); G06F 11/3034 (2013.01); G06F 2009/45591 (2013.01); G06F 2009/45595 (2013.01)] 10 Claims
OG exemplary drawing
 
1. A cluster system, comprising:
a plurality of nodes to be connected by a first network, each of the plurality of nodes including a processor and a memory; and
a storage device including a processor and a memory,
wherein the plurality of nodes include a master node, a first node performing an I/O request with respect to a first volume of the storage device capable of including a plurality of volumes through a second network, and a second node taking over processing of the first node at a time of failure of the first node,
when a failure of the first network occurs in the first node, the master node transmits a connection release instruction for releasing connection between the first node and the first volume to the storage device through the second network,
when the connection release instruction is received, the storage device releases the connection between the first node and the first volume, and updates connection state management information for managing a connection state between the first node and the first volume, and
the first node acquires the connection state to be managed by the connection state management information of the storage device through the second network, and executes post-processing relevant to the first volume that is associated with the connection release instruction, on a basis of the acquired connection state.