In an HACMP configuration, you can migrate by creating a partitioned
system, which allows you to migrate some of the clients and one of
each server pair to Blueworx Voice Response Version 6.1, while
the remaining clients and servers remain in a working Blueworx Voice Response Version 4.2 system
(although the capacity of this system will be limited).
In this configuration, you can then stop the Blueworx Voice Response Version 4.2 partition
and start the Blueworx Voice Response Version 6.1 partition.
In this way, the downtime is reduced to the time it takes to quiesce
the Version 4.2 partition,
migrate the database and then bring the Blueworx Voice Response Version 6.1 clients
and servers into the system.
Procedure
Before you start,
backup all the machines in the SSI cluster.
In the steps below,
server A denotes the SSI server used as the
DB2® server
and server B denotes the SSI server used as the NFS server:
Note: Ensure
that /home/dtdb23in/sqllib/db2nodes.cfg contains
the HA service name of the DB2 node. If it does not, edit the file
and ensure the HA system works before continuing.
Upgrade one or more clients: On the first
client machine:
- Quiesce telephony.
- Shutdown Blueworx Voice Response.
- Upgrade to Blueworx Voice Response as
described in Software installation.
- If you are also migrating to Unified Messaging Platform Version 6.1, install
the Unified Messaging base
filesets.
Failover to server B and create a partition: The DB2 and NFS
services can be failed over to server B, so that server A can be upgraded
while the system continues working at Version 4.2.
- On server A, failover all services to server B so that server
B assumes the role of a DB2 server and NFS server.
- Install the filesets to upgrade to Blueworx Voice Response Version 6.1 on
server A, but do not attempt to migrate the database at this stage.
Failback to the Version 4.2 partition (system down): The system
must now be stopped to allow services to be failed-back to server
A and enable the database to be migrated from DB2 V8.1 to DB2 V9.5.
- Quiesce telephony on the back-level clients that are taking active
calls (these clients are still on Blueworx Voice Response Version 4.2 and
have not been upgraded yet).
- Shutdown Blueworx Voice Response on
the back-level clients.
- Edit HAstartDT on server A so that the commands to start DB2 and Blueworx Voice Response are commented
out. That is, comment out the following lines:
#/usr/bin/su -$DBINSTANCE -c db2start >/dev/console 2>&1
#su dtuser -c "/usr/lpp/dirtalk/tools/vaeinit.nox"
- Failback all services from server B to server A, so that server
A assumes the role of a DB2 server and NFS server.
At this point,
the SSI cluster is down and not taking any calls.
- On server A, log in as root and run Blueworx Voice Response migration
script by typing:
- /usr/lpp/dirtalk/tools/DTdatabase –m
This migrates the DB2 V8.1 instance and database to the DB2
V9.5 level.
Start operating using the partition and migrate the partition: The Version 6.1 partition
can now be started and operation can continue using this partition
while the Version 4.2 partition
is migrated.