惠普MC Service Guard培训.pptVIP

 乔永强 HPCS MC/ServiceGuard 培训 MC/ServiceGuard 培训安排 MC/ServiceGuard 基本原理 MC/ServiceGuard 基本操作 MC/ServiceGuard 基本故障解决 MC/ServiceGuard 基本原理 Hardware Concepts Leading idea: No single point of failure The “MC” in mc/serviceguard stands for “multiple computers” and an mc/serviceguard cluster consists of between 2 to up to 16 nodes depending on the mc/serviceguard version Hardware Concepts A supported mc/service guard cluster requires redundant hardware To protect against disk failures LVM disk mirroring or disk arrays must be used; Two disk buses are also strongly recommended and should be either configured as PV groups or PV links to protect against bus or device adapter failures. At least two LAN interfaces per node are required to protect against LAN card failures. Shared Volume Group Package Concepts Application Resource: Volume group Logical volume File system IP address Shell script Function Rule: Resource given to a package cannot be given to other packages. Relocatable IP Address Stationary IP Stationary IP addresses never change. These are normal IP addresses assigned in the same way as before. Relocatable IP Relocatable IP addresses are assigned on a per-application basis. If you want to access an application, you connect to the relocatable IP addresses. This will connect you, regardless of which system or interface it has been assigned to. Cluster Concepts 2 to 16 Nodes. Status and heartbeats are exchanged. Packages transferred after failure. Major Steps in Application Recovery After a Node Failure Node fails Cluster reformation Adoptive node takes over Fsck phase Application startup recovery Application availability The Root Volume Group This VG must not contain any HA application package LVM MirrorDisk/UX is recommended The root VG may be on a shared bus or a non-shared bus. Do not use exotic network based backup utilities such as Omniback to back up the root VG. These take a long time to get installed and working. Make_recovery/Fbackup/frecover a