(remember, LUI now has a GUI! Type in 'glui'from the command line if you prefer GUIs to command line interfaces)
To get started, you have to define the installation server machine to LUI, using the mklimm command, as in:
mklimm -n myservername -t server -i 9.117.20.33 -m 255.255.255.0
Under the covers, LUI will do things like check to see if all the services are running, and warn you if they're not, and create the LUI data base in /tftpboot/lim. LUI only supports one server per group of nodes, which defines a LUI domain.
With your server defined, it's time to define your clients, as in:
mklimm -n myclient1 -t client -i 9.117.20.31
-m 255.255.255.0 -H 000000000001 -a eth0
-d 9.117.20.33 -o thor.pok.ibm.com
Under the covers, this mklimm command creates a remote root file system for the node. The remote root is NFS mounted during the installation process to provide a root file system for the installing client. If this is the first client, /tftpboot/common is created, the common portion of the root file system. For the first and all subsequent clients, /tftpboot/(ip) address is created, a unique portion of the root FS for each client. mklimm will also create an entry in /etc/bootptab for each client node you define, and remove the entry when the node definition is deleted, or when the node completes installation.
It takes about 34 meg of space for the first client's remote root file system on the server, and about 4 meg for each additional client. Your mileage will surely vary.
NEW for 1.9! ---> The auto-reboot option <--- NEW for 1.9!
There is an attribute in the client and server definitions that controls the automatic reboot of LUI clients. If a clients reboot attribute is set to "true" it will automatically reboot after the successful completion of installation. If it is set to "false" it will not. If the installation fails the client will not reboot regardless of the value of the attribute. The client attribute has a third value, "default". If a client's reboot attribute is set to "default" it will use the value of the server's reboot attribute to determine if it should reboot or not. In this manner you can set the server's attribute to the behavior you would like the majority of your clients to observer and easily change a few to behave differently.
This attribute is settable by the -r flag of the mklimm command and also through the glui.
Bear in mind that LUI currently only manages the bootp daemon. So if you are not using bootp, you will have to ensure that the clients don't simply reboot to the network again.
NEW for 1.8! ---> The mklimcs command <--- NEW for 1.8!
There is a new LUI command, mklimcs, that will allow you to define multiple clients with one command invocation. This command depends on a couple of things:
-The IP addresses you want to use are in a contiguous range.
-The IP addresses must be resolvable to names.
-There must be an /etc/MAC.info file that contains the MAC addresses.
When you issue this command, you supply a starting ip address and netmask, an optional MAC id and count, and other optional machine parameters. The command will retrieve the MAC address from /etc/MAC.info, increment the starting ip address for each client, resolve that ip address to get the long hostname and the LUI machine name, set whatever optional parameters you provided, and finally, create or add all the clients to a machine group, if specified.
For example, to create 4 clients you might issue:
mklimcs -i 1.1.1.2 -m 255.255.255.0 -I MAC2 -c 4 -d 1.1.1.251 -G mgrp
In this example your /etc/MAC.info file would look something like this:
MAC1 000629670225
MAC2 000629670226
MAC3 000629670227
MAC4 000629670228
MAC5 000629670229
MAC6 00062967022A
MAC7 00062967022B
Notice that there are additional MAC addresses in the file. The fact that the command used -c & -I will limit it to 4 clients starting at MAC2. If you do not specify -c, the default is from the MAC id specified with -I to the end of the file. If you do not specify -I, the default is the first entry in the file. So, you can see by not specifying -I or -c, it will define all entries in the file.
The relevent portion of your /etc/hosts might look like this:
1.1.1.1 node1.pok.ibm.com
1.1.1.2 node2.pok.ibm.com
1.1.1.3 node3.pok.ibm.com
1.1.1.4 node4.pok.ibm.com
1.1.1.5 node5.pok.ibm.com
1.1.1.6 node6.pok.ibm.com
1.1.1.7 node7.pok.ibm.com
Again, there are additional entries here, since our starting ip address is 1.1.1.2 and we are defining 4 nodes, we will use 1.1.1.2 through 1.1.1.5. The clients would end up looking like this (some info deleted for brevity):
lslimm -g mgrp
client name = node2
ip address = 1.1.1.2
netmask = 255.255.255.0
MAC address = 000629670226
MAC id = MAC2
default route = 1.1.1.251
long hostname = node2.pok.ibm.com
client name = node3
ip address = 1.1.1.3
netmask = 255.255.255.0
MAC address = 000629670227
MAC id = MAC3
default route = 1.1.1.251
long hostname = node3.pok.ibm.com
...and so on...
Notice that we used the machine group name in the query.