General information about the Management Cluster

If two or more SE Servers are connected to one another, then this becomes a “Management Cluster” (also called “SE Cluster”).

A Management Cluster is configured by the customer service on customer's request and serves for both operating and managing of the participating servers.
Configuring a Management Cluster does not require any own maintenance windows or downtime.

There are several crucial requirements for the structure of a Management Cluster: a Net Unit connection between the participating SE Servers (ISL-E) must exist for communication purposes, as well as a so-called CRD with an external configuration disk (or two) must be accessible by all MUs for managing special global data.

In relation to administration and operation, all MUs of the Management Cluster are equal. This means that one can centrally administrate and operate all objects of the whole SE Server configuration on every MU.

The SE Servers remain operational as long as at least one MU is functional.
Exceptions:

• For SVP operation of an SU /390 and its correct HW display, a functional MU of the own SE Server is required.
• The operation of a BS2000 VM of an SU /390 (BS2000 console and BS2000 dialog) is possible only with a functional MU of the own SE Server.
• As the SKP functionality is always delivered by a server-local MU, there are still two MUs in an SE /390 required for the redundant SKP functionality.
• The complete administration of the BS2000  VMs is possible only with a functional MU of the own SE Server.

Note:
An inhomogeneous management cluster is a management cluster consisting of an SE server with all units (MU, HNC, SU x86) with one version (e.g. version V6.5) and another SE server with all units with another version (e.g. version V6.4).
In such a case certain restrictions have to be considered.
See for this chapter Dealing with an inhomogeneous Management Cluster.

Preparing measures in regard to global data

If the SE Servers that are already operational and have customer-specific configurations are to be connected to a Management Cluster, it needs to be taken into account that there are server-overarching “global data” on all participating SE Servers. These data can come into conflict during a merger.

As it is not easily determinable which objects of the global data would “win”, the following measures should be undertaken before connecting to the Management Cluster:

• All global objects must be made unambiguous before a Management Cluster is set up.
• Unambiguousness can mean:
  • Sameness:
    • The objects must be configured in the same way on all servers (including old attributes).
    • Example: The password for the account “admin” is set on all servers to the same value.
• Unique existence:
  • Unwanted objects are deleted.
  • Example: The mail configuration for the Alarm Management is deleted on all servers except one.
• The following global objects must be taken into account for this:
  • Account (local and LDAP) as well as its properties: passwords and password properties (only for local accounts), multi-factor authentication (MFA), individual rights, descriptions
  • Alarm Management with mail configuration, mail recipient and trap recipient
  • Configuration der private IP networks DANPRxx and MONPRxx
  • Configuration of the FC networks – the FC switches and the properties for monitoring
  • SE opartion state
  • Settings for the remote service access
  • If given, server-internal SU x86 cluster

Role concept:

• The scope of every account (regardless of its role) is always the whole Management Cluster or all objects of the Management Cluster relevant for the role.
  Limitation to one SE Server is not possible.
• The main windows Overview and Management Cluster in the category “Cluster” can be accessed in SE Manager by all users with the roles Administrator, BS2000 administrator and BS2000 operator.