Sun Cluster 3.0 12/01 Data Services Installation and Configuration Guide

Known Issues

The following issues and behaviors can occur with Sun Cluster HA for BroadVision One-To-One Enterprise.

Error creating a user when the One-To-One database fails and restarts - If the database fails when all of the BroadVision One-To-One Enterprise resources are running, you can create a new user after the database comes back online. However, only the third attempt at creating the new user will succeed. Contact BroadVision support for more information on this bug.
The One-To-One database fails and the back-end hosts fail over - If the database fails and the back-end hosts fail over before the database comes back online, the BroadVision One-To-One Enterprise resources cannot come online on any cluster node. When you successfully restart the database, start the BroadVision One-To-One Enterprise resources again.
The hosts in the startup order are offline - BroadVision One-To-One Enterprise resources must be started in a particular order. The BroadVision command bvconf bootstrap lists this order. If both of the following conditions occur, the BroadVision One-To-One Enterprise processes that are configured on the hostname in the resource group will not start.
- Any of the resources in the startup order are offline.
- You start a BroadVision One-To-One Enterprise resource that is listed after the offline resources in the startup order.
If both of these conditions occur, the resource group will come online, but the processes will not start. The probe will wait for the resource group in the startup order to come online before the probe starts the BroadVision One-To-One processes for this resource.

The One-To-One Command Center connection - To connect the Command Center to BroadVision One-To-One Enterprise servers that are configured on a cluster, try one of the following options.

Force the Dynamic Control Center (DCC) to use POOP instead of IIOP. To do so, set the value of the My Computer/HKEY_CURRENT_USER/Software/BroadVision/Dynamic Control Center/4.2/Options/Use IIOP Windows registry entry to 0.

Set the IT_LOCAL_ADDR_LIST property to include the IP addresses of all of the cluster nodes and logical hostnames that will run the orbix daemon. The following list identifies sample IP addresses to add to the bv1to1.conf file.

root host: 10.10.102.225
back-end host: 10.10.102.226
Interaction Manager host: 10.10.102.222
Interaction Manager host: 10.10.102.223

In this example, add the following line to the bv1to1.conf file, under the global export section, before the IT_DAEMON_PORT property.

IT_LOCAL_ADDR_LIST = "127.0.0.1"
               + "10.10.102.222"           
               + "10.10.102.223"           
               + "10.10.102.225"           
               + "10.10.102.226"
               ;

Note -

DCC cannot recover from failover. Contact BroadVision support for more information.

Server-port conflict - By default, the orbix daemon chooses an available port number that the IT_DAEMON_SERVER_BASE and IT_DAEMON_SERVER_RANGE properties specify for use by a server that the daemon launches. When a client attempts to connect to a server for the first time, the client asks the orbix daemon for the port number. Then the client connects to the port that the orbix daemon specifies. If failover occurs after the client asks the orbix daemon for the port number but before the client connects to that port, the client might connect to the wrong server. To protect from a server-port conflict, try one of the following options.

Configure the IT_LOCAL_SERVER_BASE property for each host so that ports that the orbix daemon assigns on different nodes will never overlap. For example, if you configure BroadVision One-To-One Enterprise servers and the Interaction Manager to run on cluster nodes A, B, and C, the bv1to1.conf file will have the following entries.

export
    ...
    IT_DAEMON_SERVER_RANGE = "200";
    ...
site bv
{
    ...
    node A {
        export IT_LOCAL_SERVER_BASE = "1300";
        ...
    }
    node B {
        export IT_LOCAL_SERVER_BASE = "1500";     # 1300 + 200
        ...
    }
    node C {
        export IT_LOCAL_SERVER_BASE = "1700";     # 1500 + 200
        ...
    }
    ...
}

Add the iiop_port parameter to each process entry in the bv1to1.conf file, and ensure that no two server-port entries conflict. The iiop_port is an undocumented parameter of the BroadVision One-To-One Enterprise server that specifies which port a server should use. For example, the following process entry defines the cntdb server on port 1305.
process cntdb { parameter iiop_port = "1305"; }
C++ CORBA servers support the iiop_port parameter. For Java servers, you must upgrade to BroadVision One-To-One Enterprise 6.0AB or later versions.

The BroadVision and Oracle resource groups fail over at the same time - If you use Oracle, and the BroadVision One-To-One Enterprise backend resource groups and the Oracle resource group fail over at the same time, some BroadVision daemons might fail to restart. These daemons will fail to restart while the Oracle database is restarting. The BroadVision One-To-One Enterprise resource will attempt to restart the failed daemons until it succeeds.