Review the following topics before performing the Hardware Management Pack installation.
If Oracle Linux Fault Management (FMA) software is not working correctly, check that the following modules and services are in the correct state:
|
To check the following services and modules, see Install the Required Linux Components Before Installing Oracle Linux FMA Software.
The installer aborts when the DISPLAY variable is set on a server running Oracle Solaris OS or Linux OS. To avoid this issue, unset the DISPLAY variable before installing Hardware Management Pack.
When launching the Oracle Hardware Management Pack Installer on a Oracle Solaris OS system, the following error might appear:
./install.bin: !: not found
You can ignore this error and the Installer should launch normally.
Before installing Oracle Hardware Management Pack on a server running Oracle Solaris OS installed with the SUNWCreq (Core System Support) metacluster, you must install SUNWxcu4 (contains POSIX df command) or set the following environment variable:
IATEMPDIR=$HOME
For a Sun Fire X4170 M2 system running Windows Server 2008 R2, install the drivers from the X4170 M2 Tools and Drivers CD before using the RAIDconfig tool. Failure to install the drivers might result in the slot information for the HDDs attached to the internal ICH10 controller to be reported incorrectly.
You can also use the Oracle Hardware Installation Assistant to install Windows Server 2008 R2 to avoid this problem.
On Oracle Solaris OS 10 servers with SUNWipmi installed, the Hardware Management Pack cannot be successfully installed. The only workaround is to stop the current install using control-c, remove the installed version of SUNWipmi, and then restart the Hardware Management Pack installation.
When using the GUI mode Oracle Hardware Management Pack installer on Oracle Linux 6, the graphical installer can not be started. This is because the libXtst.i686 package is not installed by default. Either install this package before using the GUI mode, or use the console mode.
If you choose to install the QLogic support on Linux using the Oracle Hardware Management Pack installer, the process can take a long time. To make this process more efficient, install the QLogic package manually.
During the install process, the summary screen might display
Disk Space Information (for Installation Target):
Required: 169,082,111 bytes
Available: Error!
This can be safely ignored.
When using the Solaris Automated Installer (introduced with Solaris 11) to deploy software on a server, the server's Host-to-ILOM interconnect feature (required for many Oracle Hardware Management Pack features) might be left in a disabled state after the Automated Installer performs a reboot during installation. If this happens, a second server reboot after the installation has completed should fix the problem.
To determine if your server was setup by the Automated Installer, type the following command:
# netadm list | grep ncp ncp Automatic online <-- Automated Installer was used ncp DefaultFixed disabled
Installation of Oracle Linux FMA software can fail when using the Linux Anaconda installer or the Oracle System Assistant assisted OS installation (which calls the Anaconda installer). The OS installation will complete successfully and the software will be installed; however, after installation and server reboot, the required Oracle Linux FMA services will not be automatically started. Consequently, none of the Oracle Linux FMA fault events will be recorded or observed on the host.
If this happens, perform the following procedure.
Complete the OS installation process, ignoring any Oracle Linux FMA software install failure messages.
After the system reboots, login as root and make a directory for the man pages.
# mkdir -p /usr/local/share/man/man1m
Create soft links to the installed man pages.
# ln -s -t /usr/local/share/man/man1m /opt/fma/share/man/man1m/fmadm.1m /opt/fma/share/man/man1m/fmdump.1m /opt/fma/share/man/man1m/fmd.1m /opt/fma/share/man/man1m/intro.1m
Enable the appropriate services.
# chkconfig --add ksyseventd.init
# chkconfig --add fmd.init
Then, start the services.
# service ksyseventd.init start
# service fmd.init start
Use the fmadm config command to ensure that all Oracle Linux FMA software components are installed and ready.
For example:
[root@testserver16 ~]# fmadm config MODULE VERSION STATUS DESCRIPTION ext-event-transport 0.2 active External FM event transport fmd-self-diagnosis 1.0 active Fault Manager Self-Diagnosis ip-transport 1.1 active IP Transport Agent mce 1.0 active Machine Check Translator sysevent-transport 1.0 active SysEvent Transport Agent syslog-msgs 1.1 active Syslog Messaging Agent
For Oracle Linux FMA to work properly with Oracle Linux 7, the mcelog service must be running in daemon mode only. However, by default, the mcelog service in Oracle Linux 7 runs with arguments: --ignorenodev --daemon --foreground. Therefore, before installing Oracle Linux FMA 2.3.1 on a system with Oracle Linux 7 you need to reconfigure the mcelog service.
Ensure the mcelog service is installed and running by typing the command:
systemctl status mcelog
If mcelog is not installed, you will see:
[root@testserver16 ~]# systemctl status mcelog mcelog.service Loaded: not-found (Reason: No such file or directory) Active: inactive (dead)
If mcelog is not installed, install mcelog using yum by typing the command:
yum install mcelog
After successful installation, proceed to the next step.
Edit the /etc/mcelog/mcelog.conf file to do the folllowing:
Uncomment the "raw=yes" entry.
Comment out the existing "memory-ce-threshold" entry and create a new one that reads "memory-ce-threshold = 3 / 72h".
Edit the /etc/mcelog/mcelog.setup file and comment out the existing "/usr/sbin/mcelog --ignorenodev --syslog --foreground" entry and make a new one that reads "/usr/sbin/mcelog --daemon".
Edit the /etc/systemd/system/multi-usr.target.wants/mcelog.service file to change the "[Service]" section from:
[Service] ExecStartPre=/etc/mcelog/mcelog.setup ExecStart=/usr/sbin/mcelog --ignorenodev --daemon --foreground StandardOutput=syslog
to:
[Service] Type=forking ExecStartPre=/etc/mcelog/mcelog.setup ExecStart=/usr/sbin/mcelog --daemon StandardOutput=syslog
Apply the changes you have made by typing the command:
systemctl daemon-reload
Restart the mcelog service by typing the command:
systemctl restart mcelog
Confirm that mcelog is running in daemon mode by typing the command:
systemctl status mcelog
You should see the output similar to:
[root@testserver16 ~]# systemctl status mcelog
mcelog.service - Machine Check Exception Logging Daemon
Loaded: loaded (/usr/lib/systemd/system/mcelog.service; enabled)
Active: active (running) since Fri 2014-10-03 12:52:13 EDT; 6s ago
Process: 3939 ExecStart=/usr/sbin/mcelog --daemon (code=exited,
status=0/SUCCESS)
Process: 3935 ExecStartPre=/etc/mcelog/mcelog.setup (code=exited,
status=0/SUCCESS)
Main PID: 3940 (mcelog)
CGroup: /system.slice/mcelog.service
|__3940 /usr/sbin/mcelog --daemon
For Oracle Solaris or Linux installations of Oracle Hardware Management Pack beginning with version 2.3.2.2, the uninstall directory path is: /opt/ssm/setup/uninstall. For versions of Oracle Hardware Management Pack earlier than 2.3.2.2, the uninstall directory path is: /opt/sun-ssm/setup/uninstall.
After running uninstall on a Solaris or Linux system, there might still be directories listed under /opt/sun-ssm. The reason for this is that when Oracle Hardware Management Pack is upgraded, the /opt/sun-ssm directory (if it exists) is retained for compatibility with other versions of Oracle Hardware Management Pack. If you are completely removing Oracle Hardware Management Pack and all of its components, these directories can be safely deleted.