Highly Available Services with DRBD and Pacemaker

• General understanding of Linux block devices, filesystems, and TCP/IP networking
• For the "Root filesystem on DRBD" topic: perfect understanding how a Linux system boots (initramfs, init, services, runlevels, ...)

The primary objective of this practical course is to introduce the Linux high-availability stack (Pacemaker, DRBD) to system administrators. During the training, each participant will build a two-machine cluster that offers a simple highly-available service (such as PostgreSQL or NFS) on a virtual IP address. Also, uses of DRBD not related to high availability are considered.

• DRBD basics
  • Theory of operation
  • Installation
  • Supported replication modes and their use cases
  • Creation of a two-node single-primary cluster
  • Manual failover
  • Recovering from split-brain
• Corosync
  • Installation and initial configuration
• Pacemaker
  • Theory of operation
  • Installation
  • Role of STONITH devices
  • Configuration of resource agents
    • DRBD
    • Mounted filesystem
    • Virtual IP address
    • LSB and systemd services
    • Groups, ordering and colocation constraints
• DRBD in Dual-Primary mode
  • Use with OCFS2
• Advanced topic: root filesystem on DRBD