More Related Content Similar to Cubrid Inside 5th Session 2 Ha Implementation Similar to Cubrid Inside 5th Session 2 Ha Implementation (20) Cubrid Inside 5th Session 2 Ha Implementation2. 목차 1. 배경 1.1 Linux Heartbeat 1.2 CUBRID HA 2. CUBRID Heartbeat 구현 2.1 개요 2.2 구조 2.3 시나리오 – Start Node – Monitor - Deregister – Change Node – Stop Node 5. resource monitoring: resources can be automatically restarted or moved to another node on failure 6. fencing mechanism to remove failed nodes from the cluster현재 resource manager는 pacemaker 프로젝트로 분리 7. 1.2 CUBRID HA Linux Heartbeat을 이용한 CUBRID HA 구성 Slave Node Master Node copylogdb copylogdb Heartbeat Heartbeat ccm ccm lrm lrm applylogdb applylogdb cub_server cub_server heartbeat start/stop/monitor start/stop/monitor/promote/demote 10. 2.1 개요 목표 및 기대 효과 CUBRID HA 동작에 필요한 heartbeat 기능을 CUBRID 내에 구현 손쉬운 CUBRID HA 구성 Heartbeat에 의한 문제 최소화 OS 의존성 제거 11. 2.2 구조 cub_master내 heartbeat 기능 구현 cub_master local resource manager cluster node manager heartbeat receiver local resource receiver cluster node entries local resource entries Heartbeat /UDP TCP or Unix Domain Socket 26. node node #1 node #n other cluster node #1 cub_master local process cub_master cub_master cub_master 2.3 시나리오 – Change Cluster Reconfigure Nodes 클러스터 멤버 노드의 설정을 변경 1. reconfigure cluster 2. reload cluster configuration 33. node node #n node #1 other cluster node #1 cub_master local process cub_master cub_master cub_master 1. leave cluster 2. heartbeat [ bye, unknown ] 2. heartbeat [ bye, unknown ] 2.3 시나리오 – Stop Node Leave Cluster Cluster 탈퇴 절차 수행