When the system is operating normally the servers should be running the following applications.

1. Primary Server (the server actively running the SQL Anywhere Server - database server engine), and Signal Processing Application (the application that talks to the receivers and handles timing events for the monitoring software package)
   

SQL Anywhere               Signal Processing 

          

The system tray should show these icons

They may not be exactly as you see them, but you should have both icons unless your Spa application is running on another server, and or your system tray may be ‘HIDING’ them under the UP arrow to the left of this image. 

2. Secondary – a server running the  MKSCluster application (Active) – these won’t show in the system tray but should be desktop applications that are active with a dbbackup window that is maintaining the MKMSDefault.log doing a live transaction backup.  – You may also have a tertiary server that would be setup and running exactly as the secondary server running the MKSClustering Application.
   

The backup transaction window will look the image below.

 

When the system is operating in this configuration, automatic failover would occur in the following order.

• 1st Failover – Secondary #1
  
• 2nd Failover – if you have a tertiary server - Secondary #2
  

All active Cluster modules will show a message indicating a disconnect from the server has occurred and will wait the user defined Timeout in seconds before attempting to locate and connect to another SQL Anywhere Sybase Server.  The Secondary #1 Server will wait 5 seconds then attempt to scan all servers in the Cluster for an active Server.  Since no server service is currently running, Secondary #1 will assume it should be the Active server.  It will start the SQL Anywhere server service and then start SPA.

During this process, the dispatcher workstations will receive a customizable message that the database server is not available, and they will begin scanning for a new server to become active and take over that role.  As soon as one of the other cluster servers takes over the active role, the dispatcher workstations will be presented with the login dialog to connect to the new server and return to the processing in the same manner as before.

Once SPA is running, it will reconnect to the receivers through the connection with the Device Master devices.

All remaining cluster modules will reconnect to the newly activated server on Secondary #1 and continue their backup and monitoring functions.

Manually Force a Failover

This process can be followed when maintenance is required on the primary server or if there is a need to switch between servers.

1. On the Primary Server Stop and Exit SPA.
   

1. If SPA is running on the desktop you can skip this step.  If SPA is not running on the desktop, mouse over the system tray icon for SPA (Cloud and Lightning Bolt Icon) and double click it.  Alternately you can just right click on SPA, select STOP, wait a few seconds, right click on it again and click exit.  Stopping SPA in this manner does not give you the visual cues as show below.  If you do it this way, skip to Step 2.
   

2. Once SPA is running visibly on the desktop, click the STOP button to force SPA to disconnect from the receivers. Right now, the traffic light shows ‘GREEN’ meaning Spa has opened the ports and is communicating with the programmed receivers. 
   

3. Wait until the SPA form shows that the STOP button now reads START as shown. (notice the traffic signal is ‘RED’ – meaning Spa is no longer talking to the receivers. 
   

4. You can now click the Exit button to close SPA.
   

2. On the Primary Server Shutdown the active SQL Anywhere Server application.
   

1. If SQL Anywhere Server is running on the desktop you can skip this step.  If the SQL Anywhere Server is not running on the desktop, mouse over the system tray icon for SQL ('SQL' Icon) and double click it.
   

2. Once the SQL Anywhere Server application is on the desktop, you can click the Shutdown button as shown.
   

3. You will be prompted with a verification question telling you that there are active connections.  You must click the 'Yes' button to force the SQL Anywhere Server to shut down as shown.
   

3. Visually verify that neither the SQL Anywhere Server nor the Signal Processing Applications exist in the system tray before proceeding.
   

All active Cluster modules will show a message indicating a disconnect from the server has occurred and will wait the user defined Timeout in seconds before attempting to locate and connect to another SQL Anywhere Sybase Server.  The Secondary #1 Server will wait 5 seconds then attempt to scan all servers in the Cluster for an active Server.  Since no server service is currently running, Secondary #1 will assume it should be the Active server.  It will start the SQL Anywhere server service and then start SPA.

During this process, the dispatcher workstations will receive a customizable message that the database server is not available, and they will begin scanning for a new server to become active and take over that role.  As soon as one of the other cluster servers takes over the active role, the dispatcher workstations will be presented with the login dialog to connect to the new server and return to the processing in the same manner as before.

Once SPA is running, it will reconnect to the receivers via TCP/IP or serial connection.

All remaining cluster modules will reconnect to the newly activated server on Secondary #1 and continue their backup and monitoring functions.

At this point any maintenance can be done to the primary server before placing it back in service.

Recovering System

To recover the system back to normal operating conditions, follow these steps.

1. Start MKSCluster on the Primary Server - To start the MKSCluster application, first check the computer's task bar to make sure it is not already running.  When verified that it is not already running, double click the icon to launch the program.
   

2. Activate the cluster application by clicking File, and then Active from the menu as shown:
   

3. Allow the backup function to complete fully, verify that the percentage completion is at 100% and shows “waiting for next page”.  Follow the Failover procedure to Shut down SPA and the server service on the Secondary #1 server and or Secondary #2 if you have a tertiary server and that is active. 
   
4. Allow enough time for the SQL Anywhere Server and the Signal Processing application to load on the Primary Server.
   
5. Once SPA is running, it will reconnect to the receivers either via TCP/IP or a serial connection.
   
6. If you have a tertiary server preform step number 2 to reactivate MKSCluster on that server.