HP XC System Software Release Notes for Version 4.0

7 System Administration, Management, and Monitoring
This chapter contains notes about system administration, management, and monitoring.
7.1 LDAP over SSL Is Not Supported
LDAP over SSL to provide Transport Layer Security services is not supported in XC Version 4.0
7.2 Issues with Kernel Dumps
Although kexec/kdump can be configured to save kernel dumps to the local disks or across the
network to a remote node, only the network dump method is supported in HP XC.
The kdump tool has issues with some of the disk controllers like Smart Array (cciss), aacraid,
and so on.
The following are known issues with kernel dumps:
• When an x86_64 system crashes, the kexec/kdump scripts compresses the core dump file
(named vmcore). Therefore, the size of a vmcore file is smaller than the total RAM size on
the crash node. Such compressed vmcore files are readable by the crash utility. However,
on Integrity systems, it was observed that the size of the vmcore files generated is almost
equal to that of the RAM on the crashed nodes.
• On some (not all) ProLiant DL145 G2 servers, the noapic kernel parameter option must be
passed to the HP XC kernel (/boot/gub/grub.conf) for the crash dumping to complete
successfully.
Without the noapic option, when a crash occurs on such nodes, the kdump kernel might
hang after displaying a message similar to following: "Kernel panic - not syncing: IO-APIC
+ timer doesn't work! Try using the 'noapic' kernel parameter".
• In a standard configuration (that is, a configuration without improved availability), the head
node is configured to act as the kdump server to collect crash dump files from all client nodes.
However, to collect dumps resulting from head node panics, you must manually configure
a separate node as a kdump server.
Contact the HP XC Support Team at [e-mail] to obtain the kdump configuration
procedure for head node panics.
• HP recommends that you exclude the crash dumps directory on a head bode from the golden
image.
If a head node is configured as a kdump server, HP recommends that you exclude the
directory used for collecting crash dump files by adding that directory to the
base_exclude_file and updgi_exclude_file exclude files.
If the dump directory is not excluded, the crash dump vmcore files (if any) in the crash
dump directory will be included in the golden image whenever the golden image is updated.
For example, if /dumps is the crash dump collecting directory, add the entry /dumps/* on
a separate line in the following exclude files:
/opt/hptc/systemimager/etc/base_exclude_file
/opt/hptc/systemimager/etc/updgi_exclude_file
Excluding the crash dump directory prevents huge vmcore files from being included in the
golden image.
7.3 Stopping Nagios Before Resetting the Database
If you need to reset the database with the reset_db command, you must stop Nagios first:
# service nagios stop
7.1 LDAP over SSL Is Not Supported 27