|
The Partition Load Manager (PLM) provides CPU and memory
resource management and monitoring across logical partitions (LPARs).
Partition Load Manager allows you to effectively use CPU and Memory
resources by allowing you to set thresholds for designated resources.
When a threshold is exceeded, Partition Load Manager can try to assign
CPU and/or Memory resources to that LPAR by using resources assigned to
other LPARs that are not being used.
PLM is an automated mechanism for utilizing the Dynamic LPAR (DLPAR)
capabilities of the HMC and requires communication with the HMC. This
means that before PLM will function, DLPAR must be functional on the
HMC. DLPAR requires communication with each LPAR via the Resource
Monitoring and Control (RMC) subsystem.
Preparation for implementation of PLM
Install and configure SSL and OpenSSH.
Verify or install the following fileset on the PLM Server and every
PLM client LPAR::
After installation of the "csm.client" file, run the following
commands to initialize the RMC subsystem:
cd /usr/sbin/rsct/install/bin
./recfgct
lssrc -a | grep rsct
From the above "lssrc" output, check to ensure "IBM.CSMAgentRM" is
running. Repeat these steps on every PLM client LPAR.
Before implementing this rest of this procedure, verify the
HMC is able to perform DLPAR functions to the client LPAR, then
continue. If the HMC is unable to perform a DLPAR, PLM will not
work.
Implementing PLM
Install the following filesets:
- plm.license
- plm.server.rte
- plm.sysmgt.websm
For setup of PLM, create .rhosts files on the server and all
clients. After PLM has been set up, you can delete the .rhosts
files.
Create SSH keys
On the PLM server, enter:
ssh-keygen -t rsa
Copy the HMC secure keys to the PLM server
scp hscroot@hmchostname:.ssh/authorized_keys2 ~/.ssh/tmp_authorized_keys2
Append the PLM server keys to the temporary key file and copy it back
to the HMC:
cat ~/.ssh/id_rsa.pub >> ~/.ssh/tmp_authorized_keys2
scp ~/.ssh/tmp_authorized_keys2 hscroot@hmchostname:.ssh/authorized_keys2
Test SSH and enable WebSM
Test SSH to the HMC. You should not be asked for a password.
ssh hscroot@hmchostname lssyscfg -r sys
On the PLM server, make sure you can run WebSM:
/usr/websm/bin/wsmserver -enable
Configure the PLM Server
On the PLM server, open WebSM and select Partition Load Manager.
Click on ghe Globals tab and enter the fully qualified hostname of
your HMC. Enter "hscroot" as the HMC user name. Enter the CEC name,
which can be obtained by running the following command on the PLM
server:
ssh hscroot@hmchostname lssyscfg -sys -F name
Select the system name that corresponds to the frame you are configuring
in the PLM server and enter this as the CEC name.
Click on the Groups tab and add the groups "dedicated" and "shared".
The maximum values should be the total amount of CPU and memory on the
frame being configured to be managed by the PLM. Click on CPU and
memory management to manage both.
Click on the partitions tab and add all the LPAR's on the frame to be
managed by the PLM. Use the fully qualified domain name as the
partition name for each LPAR.
Click on OK to create the policy file and verify it's existance on
the PLM server under "/etc/plm/policies"
From the WebSM interface of the PLM, perform the PLM setup.
NOTE: You must be logged into the PLM server through
the WebSM interface as "root" to perform this step.
Test RMC Authentication
Test RMC authentication by running the following command from the PLM
server, where "plm_client_name" is the hostname of the LPAR that will
be managed by PLM.
CT_CONTACT=<plm_client_NAME> lsrsrc IBM.LPAR
If successful, several lines of LPAR information will be printed out
instead of "Could not authenticate user".
Start the PLM Server
From the WebSM interface of the PLM server, start the PLM server.
Enter the full path file name of the policy file name. The full path
file name of the policy file will be the directory
"/etc/plm/policies" followed by the serial number of the frame.
Any alphabetic characters in the serial number must be entered in
UPPERCASE letters.
For example:
/etc/plm/policies/10F6BEE
Also enter the full path file name of a log file where the PLM will
store activity information. Several utilities are dependent upon the
information contained within the log file so it is important that this
log file be created in the correct directory with the correct name. The
log file directory is "/var/opt/plm" and the log file name is the
serial number of the frame followed by ".log". Any alphabetic characters
in the serial number must be entered in UPPERCASE
letters. For example:
/var/opt/plm/10F6BEE.log
NOTE: You may have to "touch" the logfile before
starting the PLM Server
Troubleshooting
If the PLM server does not start, check the PLM server file
"/var/ct/cfg/ctrmc.acls" to ensure the following lines are at the bottom
of the file:
IBM.LPAR
root@hmcHostname * rw
NOTE: Even though there is no access to the "root"
user on the HMC, this line should still reference "root@hmcHostname".
On the PLM client LPAR check the same file "/var/ct/cfg/ctrmc.acls"
to ensure the following lines are at the bottom of the file. Recognize
the last line of this file on a PLM client LPAR will reference the PLM
Server hostname rather than the HMC hostname:
IBM.LPAR
root@plmServerHostName * rw
If you edit, the "/var/ct/cfg/cgrmc.acls" file on the PLM server or
on a PLM client LPAR, restart the RMC subsystem on the modified
systems.
refresh -s ctrmc
Troubleshooting
If the PLM server still does not start, there is most likely an RMC
authentication problem. Begin by obtaining a list of trusted hosts
by running the following command on the PLM server:
/usr/sbin/rsct/bin/ctsvhbal
One or more identities of the PLM client LPAR should appear in this
list. If not you may need to rerun the PLM Setup. This can be
performed from the WebSM interface or from the command line on the PLM
server. The command line is:
cd /etc/plm/setup
./plmsetup <plmClientHostName> root
On the PLM client LPAR check the list of trusted hosts by running
the following command:
/usr/sbin/rsct/bin/ctsthl -l
The PLM Server host name should appear in this list. If multiple
identities exist, it is usually a good idea to remove them all and rerun
the PLM setup command on the PLM server. To remove the trusted host
identities on a PLM client LPAR, run the following command:
/usr/sbin/rsct/bin/ctsthl -d -n <hostname>
Trusted host identities can be added on the PLM server or client
LPAR's using the following command:
/usr/sbin/rsct/bin/ctsthl -a -n <hostname> -m rsa512 -p <identifier>
Where the <identifier> can be obtained by running ctsthl -l on
the opposite system to determine it's value.
Troubleshooting
One problem that was encountered with the PLM server was when using
the WebSM interface and clicking on the link labeled "Show LPAR
Statistics", a dialog window would appear filled with java errors, and
the statistics screen would not start. This was apparently due to a
formatting problem with the policy file itself. However the PLM server
will start and there are no obvious errors other than the inability to
click on the "Show LPAR Statistics" link.
The fix for this problem is to delete the Policy file and create a
new one.
|
