Configuring dual management server

This appendix describes the basic requirements, use cases, and configuration steps for the dual management server support.

Until IBM Storage Scale System 6.1.6.0, the deployment did not formally support two management server in the same cluster. Many customers had one of three environments.
  • Single P9 or BYOMS node, which controlled a single cluster and single site.
  • Single P9 or BYOMS node, which controlled a single cluster and dual sites. Both sites networked via a stretch cluster (all networks stretch between locations).
  • Dual P9 management nodes, two locations. Only the high-speed network stretched between sites. The management networks stayed local per site. Thus, each management node could only see its local hardware.

The dual management server provides a live backup management server to become available when the primary goes down. Also, dual management server allows for multiple GUI services and collectors to become active. The GUI, in particular, allows CSI applications to stay active when one management server goes down.

Requirements

Ensure that following requirements met for the dual management server support:
  • Start of changeThe upgraded version needs to be V5.1.6.1 (IBM Storage Scale System 6.1.6.0) or higher.End of change
  • Two P9 management server
  • Single cluster
  • All networks are flat, single namespace (IBM Storage Scale System stretch cluster requirements)
  • All nodes (including two management server) have unique FQDN/IPs
  • The /etc/hosts file copied from a primary to a backup management server (the same as hosts.yml)

Support statement

Only P9 management server supported (no BYOMS, no Power8, or no mixing)

Restrictions

  • This is not an HA IBM Storage Scale System but an HA GUI/ CSI.
  • Concept of primary/secondary management server.
  • If a primary management server goes down, a secondary management server can be used to install a container and takeover the primary management server role
  • Consider one management server as primary but run all essrun commands on another management server in the list. You cannot move back and fourth between primary and secondary management server (for deployment or for GUI usage).

Configuring dual management server

  1. Connect network in the backup management server as a primary management server (Mgmt/FSP/Campus).
  2. Add backup management server network hostnames to the /etc/hosts file in an existent cluster.
  1. Connect network in the backup management server as primarily management server (Mgmt/FSP/Campus).
    bkpEMS.MGMT.Network.IP backupEms.domain backupEms 
bkpEMS.HS.Network.IP backupEms-hs.domain backupEms-hs
  2. Update the backup management server from the container that is running on the primary management server.
    essrun -N backupEms config load -p
essrun -N backupEms update --precheck
essrun -N backupEms update --offline
  3. Create network bonds.
    essrun -N backupEms network --suffix=-hs
  4. Add the backup management server to the existent cluster from the container that is running in the primary management server.
    essrun -N primarilyEms cluster --add-ems backupEms --suffix=-hs
  5. Check if successful checks.
    essrun -N backupEms config check
  6. Add the backup management server to the GUI configuration.
    essrun -N backupEms gui --configure