How to Set Up Fault Tolerant SpectroSERVERs

Document ID : KB000009404
Last Modified Date : 14/02/2018
Show Technical Document Details
Introduction:

Fault Tolerance configuration is a redundancy configuration. It consists of Primary and Secondary SpectroSERVER. 

In case of the Primary SpectroSERVER down, OneClick Server switches to the Secondary SpectroSERVER automatically. With this configuration monitoring by Spectrum won't be interuppted when Primary SpectroSERVER is down.

Additional information can be found in the online documentation below.

   Setting Up Processes in a Fault-Tolerant Environment

This document describes how to set up Fault Tolerant SpectroSERVERs, as well as some points to consider when using a Fault Tolerant environment.



Background:

How to Set Up Fault Tolerant SpectroSERVERs
Fault Tolerant
Error:"Two servers on <hostname> port 0xbeef and <hostname> port 0xbeef have the same service Landscape Service Type "Landscape" and precedence 10."

Environment:
All Spectrum versions on all platforms
Instructions:

In a Fault-Tolerant configuration, two SpectroSERVERs are configured with the same landscape, but different precedence values, as shown below:

Server 1 (Primary):
LH=0x400000
Precedence=10

Server 1b (Secondary):
LH=0x400000
Precedence=20

Fault Tolerance can also be configured in a Distributed SpectroSERVER environment, as shown below:

 

Distributed Landscape

 Master Landscape

Distributed Landscape

Server 2 (Primary):
LH=0x600000
Precedence=10

Server 1 (Primary):
LH=0x400000
Precedence=10

Server 3 (Primary):
LH=0x800000
Precedence=10

Server 2b (Secondary):
LH=0x600000
Precedence=20

Server 1b (Secondary):
LH=0x400000
Precedence=20

Server 3b (Secondary):
LH=0x800000
Precedence=20

 

In this example, the Master Landscape is the ONLY server to map the landscapes of the distributed servers. Each of these three servers knows about its respective secondary, and the Master Landscape knows about all six servers.

If the master landscape's primary server (Server 1) fails, the master secondary (Server 1b) leaves standby mode and becomes active. Servers 2 and 3 will automatically communicate with Server 1b. When Server 1 goes green again, all connections will be switched back to Server 1 from Server 1b and Server 1b will return to standby mode. 

The point to keep in mind is that Landscape handles MUST be different in a distributed environment (regardless of fault tolerance), and Precedence values MUST be different in a fault tolerant environment (regardless of distributing SERVERs).

Assuming that you have built a database on a Primary server and have installed SPECTRUM on your Secondary server, follow these steps to configure your Secondary server:

  1. Do a database save on the Primary server, either with Online Backup or the  SSdbsave  from the command line (Offline Backup). 
  2. Move the saved file to the Secondary machine using FTP (make sure FTP is in binary mode) or SCP. Place the file in the  <specroot>/SS  directory on the secondary machine. 
  3. On the secondary machine, make sure the SpectroSERVER is NOT running, and then type the following command from your  <specroot>/SS  directory:

    ../SS-Tools/SSdbload -add 20 -il <filename>

    where <filename> is the name of the savefile you transferred from the primary server.

  4. On the Primary server, bring up the Online Backup window and schedule daily Online Backups. Each time Online Backup runs, it will synchronize the databases of the Primary and Secondary servers. 
  5. Start the server on the Secondary machine.

Check the demo video.

 

Points to Keep in Mind

The Secondary SERVER will not poll devices unless it loses contact with the Primary SERVER. If you want the Secondary to continually poll all models (thereby making it a hot stand-by instead of a warm stand-by), you need to add the following line to the  $SPECROOT/SS/.vnmrc  file on your Primary SpectroSERVER:


secondary_polling=yes

 

Additional Information: