If a DB2® pureScale® instance is
already installed, follow these instructions to apply a new fix pack
level. If a DB2 pureScale instance is
not already installed, use the instructions in Installing the DB2 pureScale Feature using the DB2 Setup wizard (AIX and Linux).
Before you begin
- Ensure that you meet all of the requirements before you install
a fix pack. Refer to Preparing to install a fix pack.
- Ensure
that you are logged in as root.
- Issue db2stop
command.
Procedure
To update an existing DB2 pureScale instance to a new
fix pack level, follow these instructions.
- Determine the level of SA MP on your installation
and on the installation media.
- Query the SA MP Base Component level already installed
on your system:
<image_path>/db2/<platform>/tsamp/db2cktsa -v install
- Query the level of SA MP Base Component on the installation
media:
<image_path>/db2/<platform>/tsamp/db2cktsa -v media
- Determine the level of IBM General Parallel File System
( GPFS™) installed:
- Query the level if GPFS already
installed on your system:
<image_path>/db2/<platform>/gpfs/db2ckgpfs -v install
- Query the level of GPFS on
the installation media:
<image_path>/db2/<platform>/gpfs/db2ckgpfs -v media
- As an instance user, you must :
- Stop the database manager on all hosts by issuing the
following command on a single host:
db2stop
- On the same host that you issued db2stop from,
you must stop the DB2 instance
for each host in the cluster. You can stop each instance by issuing
the following command:
db2stop instance on hostname
where hostname is
the host name for a member or CF. This command must be issued for
each host in the cluster. For example, if your cluster consists of host1, host2,
and host3, you must issue the db2stop command
three times.
- If either the level of SA MP Base Component or the level of GPFS software on the installation
media is higher than the level already installed on your system, the DB2 cluster services administrator
must place all hosts in the cluster into maintenance mode. The maintenance
mode must be cluster manager (CM) maintenance mode.
To
place the cluster into CM maintenance mode, as a DB2 cluster services administrator, enter the db2cluster command
on any host in the cluster:
DB2DIR/bin/db2cluster -cm -enter -maintenance -all
where DB2DIR represents the installation
location of your DB2 copy.
- If the level of GPFS on the installation media is higher than
the level already installed on your system, the DB2 cluster services administrator must place
all hosts in the cluster into maintenance mode. The maintenance mode
must be cluster file system (CFS) maintenance mode. To place a cluster
into CFS maintenance mode, as a DB2 cluster
services administrator, enter the db2cluster command
on any host in the cluster:
DB2DIR/bin/db2cluster -cfs -enter -maintenance -all
where DB2DIR represents
the installation location of your DB2 copy.
- Update each host in the DB2 pureScale instance to Version 10.1 Fix
Pack 3 (or higher) on each host. For example, to update to Fix Pack
3, run the installFixPack command on each host:
installFixPack -b /opt/fp2level/path -f level -p /opt/fp3level/path -t /tmp/trace-file-path -L -l /tmp/install-log-path
The
path specified in the -p parameter must be different
than the path specified in the -b parameter.
- If the DB2 cluster services administrator placed the
cluster into CM maintenance mode, the DB2 cluster
services administrator must remove the cluster from CM maintenance
mode. To remove the cluster from CM maintenance mode, enter the following
commands on each host in the DB2 pureScale environment:
DB2DIR/bin/db2cluster -cm -exit -maintenance
DB2DIR/bin/db2cluster -cfs -repair -network_resiliency -all
where DB2DIR represents
the installation location of your DB2 copy.
- If the DB2 cluster services administrator placed the
cluster into CFS maintenance mode, the DB2 cluster
services administrator must remove the cluster from CFS maintenance
mode. The cluster cannot be in CM maintenance mode at this time. To
remove the GPFS cluster from
CFS maintenance mode, enter the db2cluster command
on each host in the DB2 pureScale environment as
the DB2 cluster services administrator:
DB2DIR/bin/db2cluster -cfs -exit -maintenance
where DB2DIR represents
the installation location of your DB2 copy.
- Update the instance to the new installation path using the db2iupdate command.
However, in a geographically dispersed DB2 pureScale cluster (GDPC) environment, the
db2iupdt command
issues a warning indicating that the
db2cluster command
failed (
db2cluster -cfs -verify -configurations).
To avoid this warning, before running the
db2iupdt command,
enter:
root@hostA1>export PASS_USER_MANAGED_GPFS_VALIDATION=YES
On
each host, update the instance to the new installation path:
DB2DIR/instance/db2iupdt <InstName>
Where
DB2DIR represents
the new fix pack directory specified in a previous step, and
InstName specifies
the name of the instance.
- The DB2 cluster services
administrator must commit the updates made to DB2 cluster services and make them available
to the DB2 database system.
On one host, run the db2cluster commands:
DB2DIR/bin/db2cluster -cfs -commit
DB2DIR/bin/db2cluster -cm -commit
where DB2DIR represents
the new fix pack level installation location of your DB2 copy.
- Determine the host failure detection time for the current RSCT
communication group. To retrieve the detection time:
db2cluster -cm -list -HostFailureDetectionTime
- Start the DB2 instance
for each host in the cluster by issuing the db2start command:
db2start instance on hostname
where hostname is
the host name for a member or CF. This command must be issued for
each host in the cluster. For example, if your cluster consists of host1, host2,
and host3, you must issue the db2start command
three times.
To update the cluster manager resource model, issue
the following command:
DB2DIR/bin/db2cluster -cm -repair -resources
where DB2DIR represents the new
fix pack level installation location of your DB2 copy.
- If you are planning to use multiple communication adapter ports,
install and configure additional adapters, switches and cables.
- Configure RSCT communication group host failure detection time
for the new cluster interconnect netname. As root, change the host
failure detection time:
db2cluster -cm -set -option HostFailureDetectionTime -value <value>
where value is
the detection time retrieved previously.
- If you are planning to use multiple communication adapter ports,
update the member or CF servers to use multiple netnames:
db2iupdt -update -cf <CFHostName> -cfnet CFnetname1,CFnetname2 <InstName>
db2iupdt -update -m <MemberHostname> -mnet MemberNetname1,MemberNetname2 <InstName>
- As the instance owner, refresh the resource model:
db2cluster –cm -repair –resources
- If the hosts were placed into maintenance
mode, the DB2 cluster services
administrator must start the DB2 instance
on all hosts. To start the DB2 instance
run the db2start command on each host in the DB2 pureScale environment
db2start instance on <hostname>
where <hostname> represents
the host name for a given member or CF, and the db2start instance
on hostname command is run for each host
in the cluster.
- To start the database manager, run the db2start command:
su - <InstName>
db2start
exit
where <Instname> represents
the instance owner name.
- To uninstall the old DB2 copy,
run the db2_deinstall command:
DB2DIR/install/db2_deinstall -a
where DB2DIR is
the installation path of the old DB2 copy.