The high availability manager supports multihomed hosts,
which means that the product processes can communicate with each other
even if they are running on different versions of the product. If
you are running Version 6.0.2 and later processes that need to communicate
with each other, you need to consider there are high availability
manager-related issues.
About this task
When processes are created on Version 6.0.2 or later nodes,
the host name field in the DCS_UNICAST_ADDRESS endpoint is set to
an asterisk. When you use an asterisk in the host name field, multihome
support allows the high availability manager to open and accept connections
using any IP address that is valid for that machine.
Special considerations
are required when Version 6.0.2 nodes interoperate with Version 6.0
and Version 6.0.1 nodes without Fix Pack 2 (6.0.1.2), because these
earlier versions do not contain high availability manager multihome
support. Version 6.0 and Version 6.0.1 processes do not recognize
an asterisk as a valid value in the DCS_UNICAST_ADDRESS host name
field. Therefore, these processes cannot connect to Version 6.0.2
processes that are configured with an asterisk in the DCS_UNICAST_ADDRESS
host name field. When Version 6.0.2 processes must interoperate with
Version 6.0 or Version 6.0.1 processes, the DCS_UNICAST_ADDRESS for
all Version 6.0.2 processes must be configured to use a value other
than an asterisk. This configuration is done by setting the host name
field in the DCS_UNICAST_ADDRESS endpoint to a host name or to a textual
IP address.
- If the host is configured to use a single IP address, a string
host name, for example, myhost.mydomain, is sufficient.
- If the host is configured to use multiple IP addresses, then a
textual IP address, for example, 192.168.0.2, is required.
As you add Version 6.0.2 nodes and create servers into
a mixed-release cell, you must set the host name field of the DCS_UNICAST_ADDRESS
for all processes on the new nodes, including the node agent, to one
of the two values previously specified. You must then restart the
Version 6.0.2 processes to pick up the new value.
To change
the host name field for a process:
Procedure
- In the administrative console, go to the configuration
data for one of the processes on the new nodes:
- For an application server, click server_name. Then in the Additional
Properties section, click > Ports > port_name .
- For a node agent, click node_name. Then in the Additional Properties section, click > Ports > port_name .
- For a deployment agent, click . Then in the Additional
Properties section, click > Ports > port_name .
- Click View associated transports for the port that
is associated with the DCS transport channel you want to review.
- Click DCS_UNICAST_ADDRESS, and enter the name of
the new host in the Host field.
- Click OK, and then click Save.
- Repeat the previous steps until the new host name is added
for all of the processes on the new nodes.
- Select Synchronize changes with nodes, and then
click Save again.
- Stop all of the servers that contain the processes with
configuration changes.
- Restart these servers.
Results
All of the processes can communicate with each other.
What to do next
If Version 6.0.2 processes are not configured properly, the
Version 6.0 and Version 6.0.1 processes might not start or might not
be able to connect to the Version 6.0.2 processes.
The condition
can be detected in one of the following ways:
- Version 6.0 and Version 6.0.1 processes log the following message:
HMGR0024W: An error was encountered while looking up the IP address for
the host name of a core group member. The host name is * and the server
name is myCell01\myCellManager01\dmgr. The member is excluded from the
core group.
This message indicates that the host name
for another process in the core group is an asterisk.
- The Version 6.0, Version 6.0.1, and Version 6.0.2 processes form
separate views if the Version 6.0 and Version 6.0.1 processes do not
connect to Version 6.0.2 processes. To detect these views, examine
the HMGR0218 messages that the various processes logged. If the processes
connect, the same message is logged across all processes. Specifically,
processes that are connected have the same view identifier and the
same number of processes in the view.
HMGR0218I: A new core group view is installed. The core group is
defaultCoreGroup. The view identifier is (8:0.spoletoCell01\
spoletoCellManager01\dmgr). The number of members in the new view is 2.