Posts tagged ·
·...
This PDF takes you through a sample configuration setup using DB2 UDB –
This technote describes how to create a DB2® managed GPFS™ cluster using the db2cluster_prepare utility and how to clean the GPFS cluster manually from the DB2 pureScale™ environment.
Before you begin
About this task
In a DB2 pureScale environment, all data and logs must be on an IBM General Parallel File System (GPFS). You may need to create a DB2 managed GPFS cluster and file system. This can be done without an instance by using the db2cluster_prepare command.
Procedure
Perform these steps to create a DB2 managed GPFS cluster and file system on a host which has no previously existing GPFS cluster. This is the recommended method to setup a DB2 pureScale environment.
a. Use the db2cluster_prepare command to create a DB2 managed GPFS.
db2cluster_prepare -instance_shared_dev /dev/hdisk1
b. Verify the GPFS cluster exists using the following commands. For example:
/opt/IBM/db2/V9.8/instance/db2cluster –cfs –list –domain
/opt/IBM/db2/V9.8/bin/db2greg -dump
Example output:
V,GPFS_CLUSTER,NAME,db2cluster_20110331115346.in.ibm.com,-,DB2_CREATED
a. The db2greg command will fail if you already have a GPFS_CLUSTER variable record
in the global registry. You can remove the existing record using the following command:
db2greg -delvarrec service=GPFS_CLUSTER
b. Clean the GPFS from the system using the instructions in tech note 1447939
How can you setup HADR with TSA that would use Virtual IP in IBM DB2 for Linux- UNIX and Windows?
The private and the public networks need to be setup correctly beforehand.
The private network will be used for the communication between primary and standby node for the log replay.
The public network will be used for the applications connecting to the primary and standby (which will become primary after failover) nodes.
A virtual IP address needs to be created which does not need to be attached/mapped to NIC, as this is a virtual IP(VIP). This IP just needs to be provided during the setup through db2haicu.
During the failover, the VIP will be automatically mapped to the public network (eth0 or en0) on the working node and the clients will be automatically re-routed through this IP to use the other node.
The public network IP and hosts need to be defined in /etc/hosts file on both primary and standby servers.
The HADR_LOCAL_HOST and HADR_REMOTE_HOST database configuration parameters need to be assigned with values assigned for the private network.
The port number assigned to SVCENAME dbm cfg parameter needs to be correctly defined in /etc/services file and it should not be used by any other application.
This port will be used for VIP routing.
The VIP is also used for ACR (Automatic Client Re-route).
Here is the link to the white paper which describes in detail how to setup HADR with VIP:
http://download.boulder.ibm.com/ibmdl/pub/software/dw/data/dm-0908hadrdb2haicu/HADR_db2haicu.pdf
