Configuring proxy servers for failover

The proxy server failover setup requires the Tivoli Netcool/OMNIbus basic failover architecture, and the following additional components: a primary proxy server and a backup proxy server.

About this task

The following figure shows the configuration for proxy server failover.

Figure 1. Proxy server failover setup
This figure shows how to set up proxy servers in a failover configuration.

In the basic failover setup, alert data from the primary aggregation ObjectServer is replicated in the backup aggregation ObjectServer through a bidirectional ObjectServer Gateway. If a connection to the primary aggregation ObjectServer fails, the clients attempt to connect to the backup aggregation ObjectServer. As shown in the figure, you must set up a virtual proxy server pair to which probes can connect. Set up the primary proxy server PROXY_P to have a single connection to the primary aggregation ObjectServer AGG_P. Set up the backup proxy server PROXY_B for failover by configuring PROXY_B to connect to the backup aggregation ObjectServer AGG_B.

Proxy servers should not be pointed at virtual ObjectServers for two reasons:
  • Proxy servers cannot fail-over and have no fail-back functionality, which can lead to unwanted situations where probes are left connected to the backup ObjectServer even though the primary proxy server and primary ObjectServer are available.
  • If the ObjectServers are using SSL, proxy servers cannot connect to failover ObjectServers. When making an SSL connection to an ObjectServer, the proxy server expects the Common Name of the certificate that is presented by the ObjectServer to be the same as the ObjectServer name itself, which will not be true when connecting to a virtual ObjectServer pair. Proxy servers lack the ability to over-ride this expectation for this reason. The virtualisation is best done at the proxy server level. Either enable the triggers to automatically disconnect probe connections from the backup ObjectServer, when a bi-directional gateway between the ObjectServers has finished its resync, or configure the probes to poll the primary server, if the probe detects it is connected to a backup ObjectServer, and then fail-back to the primary if it detects the primary is back up again.

Procedure

Using the architecture shown in the preceding figure, configure the proxy servers for failover as follows:

  1. Sset up the server communications details in the connections data file ($NCHOME/etc/omni.dat or %NCHOME%\ini\sql.ini).
    Use the following sample configuration as a guideline:
    [AGG_P]
    {
            Primary:        nchost1 10000
    }
    [AGG_B]
    {
            Primary:        nchost2 10001
    }
    [AGG_GATE]
    {
            Primary:        nchost2 10002
    }
    [AGG_V]
    {
            Primary:        nchost1 10000
            Backup:         nchost2 10001
    
    }
    [PROXY_P]
    {
            Primary:        nchost1 10003
    }
    [PROXY_B]
    {
            Primary:        nchost2 10004
    }
    [PROXY_V]
    {
            Primary:        nchost1 10003
            Backup:         nchost2 10004
    }
  2. Configure probes to connect to the proxy servers. In the probe properties file:
    • Set Server to PROXY_P.
    • Set ServerBackup to PROXY_B.
  3. In the primary proxy server (PROXY_P) properties file, set the RemoteServer property to AGG_P, as shown in the sample configuration example.
  4. In the backup proxy server (PROXY_B) properties file, set RemoteServer to AGG_B, as shown in the sample configuration example.

Results

With the sample configuration, when AGG_P fails, probes attempting to connect to PROXY_P will fail to connect and then fail-over to PROXY_B, which will in turn connect to AGG_B.