Before you begin the installation, be sure that you have the following
items and information:
__ 1.
| Ensure that your system meets all of the memory, hardware, and software
requirements to install your DB2 product. For more information, see Planning for Installation.
|
__ 2.
| The hostname of the Control workstation and all the workstations that will
participate in your partitioned database system. To resolve a
workstation's hostname, log on to the workstation and enter the following
command:
hostname
Make note of this output for each workstation.
![Figure car not displayed.](car.gif)
|
If you are planning to use every RS/6000 SP workstation in your partitioned
database system, skip this item and proceed to the next requirement.
|
|
__ 3.
| Have a file that lists the hostnames for all of the RS/6000 SP
workstations that will participate in your partitioned database system.
Set up the working collective to distribute commands to this list of
workstations by performing the following steps:
Step a.
| Log on to the Control workstation as a user with root authority
| Step b.
| Create a file called eeelist.txt that will list the
hostnames for all of the workstations that you want to install this
product on.
![Figure hint not displayed.](hint.gif)
| To resolve a workstation's hostname, enter the hostname
command.
|
For example, assume that you wanted to install this product on two SP nodes
called workstation1 and workstation2. The
contents of this file would be:
workstation1
workstation2
| Step c.
| Update the working collective. To update this list, enter the
following command:
export WCOLL=/eeelist.txt
where eeelist.txt is the name of the file that you created
that lists the RS/6000 SP workstations where you want to install DB2
EEE.
| Step d.
| Verify that the names in the working collective are indeed the
workstations that you want to install this product on by entering the
following command:
dsh -q
You will receive output similar to the following:
Working collective file /eeelist.txt:
workstation1
workstation2
Fanout: 64
|
|
__ 4.
| Verify that Network File System (NFS) is running on each workstation that
will participate in your partitioned database system by entering the following
command:
lssrc -g nfs
The Status field for each process should indicate
active.
|
__ 5.
| Ensure that the rpc.statd and
rpc.lockd processes are active on each workstation that will
participate in your partitioned database system by entering the following
command:
ps -ef | grep rpc
|
__ 6.
| Have a file system that is available to all of the RS/6000 SP workstations
that will participate in your partitioned database system. This file
system will be the home file system for the instance owner and the
Administration Server. To create, NFS export, and NFS mount this file
system, perform the following steps:
Step a.
| Log on to a workstation that will participate in your partitioned database
system as a user with root authority. Typically, this is
would be a dedicated RS/6000 SP workstation where the home directory of your
DB2 users will reside.
| Step b.
| Create a home file system for your partitioned database system called
/home2, by performing the following steps:
![Figure hint not displayed.](hint.gif)
| In our examples, we will assume that you created a file system called
home2 and its mount point is /home2.
|
- Enter the smit jfs command.
- Click on the Add a Journaled File System icon.
- Click on the Add a Standard Journaled File System icon.
- Select a volume group from the Volume Group Name list where you
want this file system to physically reside.
- Set the SIZE of file system (in 512-byte blocks)
(Num.) field to 180 000 (this is about 90
MB).
- Enter a mount point for this file system in the MOUNT POINT
field. For example, /home2.
- Set the Mount AUTOMATICALLY at system restart field to
yes.
The remaining fields can be left to the default settings. For more
information, refer to your AIX Administration Guide.
- Click on OK.
| Step c.
| Mount the file system that you just created by entering the following
command:
mount /home2
![Figure hint not displayed.](hint.gif)
|
You should not mount this file system using the AMD or
automounter utilities. These utilities can cause NFS
mounting or locking problems in a partitioned database system.
|
| Step d.
| NFS export the /home2 file system so that it is available to
all of the RS/6000 SP workstations that will participate in your partitioned
database system by performing the following steps:
- Enter the smit nfs command.
- Click on the Network File System (NFS) icon.
- Click on the Add a Directory to Exports List icon.
- Enter the pathname and directory to export (for example,
/home2) in the PATHNAME of directory to export
field.
- Enter the name of each workstation that will participate in your
partitioned database system in the HOSTS allowed root access
field. You recorded these hostnames in Before You Begin.
![Figure hint not displayed.](hint.gif)
| If you are using a high speed interconnect, we recommend that you specify
the high speed interconnect names for each workstation is this field as
well.
|
The remaining fields can be left to the default settings. For more
information, refer to your AIX Administration Guide.
- Click on OK.
| Step e.
| Log out.
| Step f.
| Log on to each workstation that will participate in your
partitioned database system and NFS mount the file system that you exported by
performing the following steps:
- Enter the smit nfs command.
- Click on the Network File System (NFS) icon.
- Click on the Add a File System for Mounting icon.
- Enter the pathname of the mount point in the PATHNAME of the mount
point (Path) field.
The path name of the mount point is where you should create the home
directory for the instance owner and the Administration Server. For
example, /home2.
- Enter the pathname of the remote directory in the PATHNAME of the
remote directory field.
For our example, you should enter the same value that you entered in the
PATHNAME of the mount point (Path) field.
- Enter the hostname of the machine where you exported the file
system in the HOST where the remote directory resides field.
This is the hostname of the machine where the file system that you are
mounting was created.
![Figure hint not displayed.](hint.gif)
| To improve performance, you may want to NFS mount the file system that
you created over a high speed interconnect. If you want to mount this
file system using a high speed interconnect, you must enter its name in the
HOST where remote directory resides field.
You should be aware that if the high speed interconnect ever becomes
unavailable for some reason, every workstation that participates in your
partitioned database system will lose access to these user's home
directory.
|
- Set the MOUNT now, add entry to /etc/filesystems or both? field
to both.
- Set the /etc/filesystems entry will mount the directory on system
RESTART field to yes.
- Set the MODE for this NFS file system field to
read-write.
- Set the Mount file system soft or hard field to
soft.
A soft mount means that the machine will not try for an infinite
period of time to remotely mount the directory. A hard mount means that
your machine will infinitely try to mount the directory, this could cause
problems in the event of a system crash. We recommend that you set this
field to soft.
The remaining fields can be left to the default settings. For more
information, refer to your AIX Administration Guide.
- Ensure that this file system is mounted with the Allow execution of
SUID and sgid programs in this file system? field set to
Yes. This is the default setting.
- Click on OK.
| Step g.
| Log out.
|
|
__ 7.
| Create three separate groups and user accounts for the:
- DB2 instance owner
- user that will execute fenced UDFs (user defined functions) or stored
procedures
- Administration Server.
The usernames you create must conform to both your operating system's
naming rules, and those of DB2. For more information on naming rules,
see Appendix D, Naming Rules.
To create all three of these users, perform the following steps:
Step a.
| Log on to the Control workstation as a user with root
authority.
![Figure hint not displayed.](hint.gif)
| If you are using NIS and file collections is enabled, you should set up
your system to exclude the propagation of any user administration
files. For more information, refer to your NIS documentation.
|
| Step b.
| Create a group for the instance owner (for example, db2iadm1),
the user that will execute UDFs or stored procedures (for example,
db2fadm1), and the Administration Server (for example,
db2asgrp) by entering the following commands:
mkgroup id=999 db2iadm1
mkgroup id=998 db2fadm1
mkgroup id=997 db2asgrp
| Step c.
| Create a user that belongs to each group that you created in the previous
step by entering the following commands:
mkuser id=1004 pgrp=db2iadm1 groups=db2iadm1 home=/home2/db2inst1 db2inst1
mkuser id=1003 pgrp=db2fadm1 groups=db2fadm1 home=/home2/db2fenc1 db2fenc1
mkuser id=1002 pgrp=db2asgrp groups=db2asgrp home=/home2/db2as db2as
| Step d.
| Set an initial password for each user that you created by entering the
following commands:
passwd db2inst1
passwd db2fenc1
passwd db2as
| Step e.
| Log out.
| Step f.
| Log on to the Control workstation as each user that you created
(db2inst1, db2fenc1, and db2as). You
will be prompted to change each user's password since this is the first
time that these users have logged onto the system.
| Step g.
| Log out.
| Step h.
| Log on to the Control Workstation as a user with root
authority.
| Step i.
| Distribute each group and user account that you created to the other SP
workstations that will participate in your partitioned database system by
entering the following command:
dsh /var/sysman/supper update user.admin
![Figure hint not displayed.](hint.gif)
| If you are installing DB2 EEE on every RS/6000 SP workstation in your
environment, you can specify the -a flag as follows:
dsh -a /var/sysman/supper update user.admin
|
This command will automatically be distributed to each RS/6000 SP
workstation that you listed in the eeelist.txt file.
![Figure hint not displayed.](hint.gif)
| If you do not issue the dsh command, the cron job
will automatically create these groups and user accounts when it is
automatically run. The default setting is 10 minutes past the
hour.
|
| Step j.
| Log out.
|
|