2-3. Goldilocks Cluster 주요 기능
        1) replica
        2) shard
        3) failover
        4) online scale out
        5) (etc) global index, sequence


3) Cluster failover

Cluster group 내의 모든 멤버는 같은 데이터를 갖지만, Shared Nothing Disk 구조이기 때문에 노드 장애 시 다른 노드를 통해 지속적인 운영이 가능하다.  
즉 1by2로 구성된 클러스터에서 G1N1 장비에 장애가 발생하더라도, 동일한 데이터를 갖는 G1N2로 failover가 진행된다.

Failover는 Cluster의 Failover Thread를 통해 진행되며, 특정 노드나 네트워크 장애 시 Member에 대한 Offline 및 Coordinator 재선정 등의 Failover가 처리된다.



  • G1N1 Shutdown 시
[2019-01-29 13:33:12.569597 INSTANCE(GOLDILOCKS) THREAD(25991,139630611420928)] [WARNING]
[CDISPATCHER-0] finalize receiver socket(member pos=0), last errno=11, by discon command(0)
(GOLDILOCKS) ERR-HY000(58010): broken socket

[2019-01-29 13:33:12.569592 INSTANCE(GOLDILOCKS) THREAD(25991,139630517151488)] [WARNING]
[CDISPATCHER-1] finalize receiver socket(member pos=0), last errno=11, by discon command(0)
(GOLDILOCKS) ERR-HY000(58010): broken socket

[2019-01-29 13:33:12.669908 INSTANCE(GOLDILOCKS) THREAD(25991,139630611420928)] [WARNING]
[CDISPATCHER] set connection(target member pos=0); physical connection(0), consensus(1)
(GOLDILOCKS) ERR-HY000(58010): broken socket

[2019-01-29 13:33:12.669944 INSTANCE(GOLDILOCKS) THREAD(25991,139630611420928)] [INFORMATION]
[CDISPATCHER-0] finalize sender socket(member pos=0)

[2019-01-29 13:33:12.669935 INSTANCE(GOLDILOCKS) THREAD(25940,140005945087744)] [INFORMATION]
[FAILOVER] begin - failover member(0)

[2019-01-29 13:33:12.670525 INSTANCE(GOLDILOCKS) THREAD(25940,140005945087744)] [INFORMATION]
[FAILOVER] acquire failover lock

[2019-01-29 13:33:12.670605 INSTANCE(GOLDILOCKS) THREAD(25940,140005945087744)] [INFORMATION]
[FAILOVER] domain coordinator failover - all members in group (1) are inactive.

[2019-01-29 13:33:12.670617 INSTANCE(GOLDILOCKS) THREAD(25940,140005945087744)] [INFORMATION]
[FAILOVER] global scn of member(1) is ‘2,0,7'

[2019-01-29 13:33:12.670623 INSTANCE(GOLDILOCKS) THREAD(25940,140005945087744)] [INFORMATION]
[FAILOVER] global coordinator failover - new global coordinator member(1)

[2019-01-29 13:33:12.670661 INSTANCE(GOLDILOCKS) THREAD(25940,140005945087744)] [INFORMATION]
disconnct member - local member pos(1), forward member pos(1), target member pos(0)

[2019-01-29 13:33:12.706747 INSTANCE(GOLDILOCKS) THREAD(25991,139630498268928)] [INFORMATION]
[CDISPATCHER-0] disconnecting member - driver member pos (1), target member pos (0)

[2019-01-29 13:33:12.806889 INSTANCE(GOLDILOCKS) THREAD(25991,139630498268928)] [WARNING]
[CDISPATCHER] set connection(target member pos=0); physical connection(0), logical connection(0), consensus(0)

[2019-01-29 13:33:12.806904 INSTANCE(GOLDILOCKS) THREAD(25991,139630498268928)] [WARNING]
[CDISPATCHER] set connection(target member pos=0); physical connection(0), logical connection(0), consensus(0)

[2019-01-29 13:33:12.806948 INSTANCE(GOLDILOCKS) THREAD(25991,139630498268928)] [INFORMATION]
[CDISPATCHER-1] finalize sender socket(member pos=0)

[2019-01-29 13:33:12.907094 INSTANCE(GOLDILOCKS) THREAD(25940,140005945087744)] [INFORMATION]
[FAILOVER] member(0) has failovered

[2019-01-29 13:33:12.907136 INSTANCE(GOLDILOCKS) THREAD(25940,140005945087744)] [INFORMATION]
[COORDINATOR] failover domain coordinator - all members in group (1) are inactive

[2019-01-29 13:33:12.907155 INSTANCE(GOLDILOCKS) THREAD(25940,140005945087744)] [WARNING]
reset remote session map - member(0)

[2019-01-29 13:33:12.907296 INSTANCE(GOLDILOCKS) THREAD(25940,140005945087744)] [INFORMATION]
[FAILOVER] finished

[2019-01-29 13:33:12.993033 INSTANCE(GOLDILOCKS) THREAD(25940,140006861182720)] [INFORMATION]
[FAILOVER] offline member - member(0)

[2019-01-29 13:33:12.993087 INSTANCE(GOLDILOCKS) THREAD(25940,140006861182720)] [INFORMATION]
offline member - member(0)

[2019-01-29 13:33:12.993623 INSTANCE(GOLDILOCKS) THREAD(25940,140006861182720)] [INFORMATION]
end offline member

[2019-01-29 13:33:12.994138 INSTANCE(GOLDILOCKS) THREAD(25940,140006861182720)] [INFORMATION]
[FAILOVER] end offline member



gSQL> select * from x$cluster_connection;

ERR-HY000(16354): connection of member 'G1N1' is broken




Networks 단절

CLUSTER_HEARTBEAT_INTERVAL
CLUSTER_HEARTBEAT_RETRY_COUNT


'IT Engineer > Database' 카테고리의 다른 글

cluster replica  (0) 2019.02.13
cluster shard  (0) 2019.02.13
Global secondary index  (0) 2019.02.13
global sequence  (0) 2019.02.13
Application단에서의 AUTO-COMMIT ON/OFF 여부에 따른 성능 차이  (2) 2018.09.19

+ Recent posts