 |
» |
|
|
|
Proper maintenance of the file system is crucial to the operation of the HP XC system. This section describes key directories and offers guidelines for the maintenance of the file system. The basic file system layout is the same as that of a standard Linux file system. Additions to the layout for the HP XC system are described in this section. Figure 1-1 illustrates the top of the file system hierarchy for a node in the HP XC system. | | | |  | IMPORTANT: The HP XC system relies on key files. Interfering with these files can cause the system to fail. The best way to avoid this situation is to respect the placement of directories and files, especially when installing software packages. The file system layout is structured to isolate the files specific to the HP XC System Software from base operating system files. This isolation enables flexibility and ease of potential upgrades of the system software and keeps software from conflicting with user-installed software. You can avoid many pitfalls by strictly maintaining separation between the HP XC System Software and other software. | | | | |
The following sections describe the base operating system files, directories specific to the HP XC System Software, system software configuration files, and log files. Key Operating System Directories | |
Following are the key directories: - /bin
Holds the binary executable files that any user can invoke. - /sbin
Holds the binary executable files that are specifically for the system administrator or designated privileged users. - /etc
Contains files for the configuration of the system and components of the system, including networking information, printers, and so on. - /hptc_cluster
Reserved for the exclusive use of the HP XC System Software. This directory is a file system that is mounted on all the nodes in the HP XC system. This systemwide directory contains key directories and files for global system use. See “Systemwide Directory, /hptc_cluster” for more information. - /opt
Holds software packages from HP and other vendors. - /opt/hp
Is reserved for optional HP applications and utilities that apply to the HP XC system. HP-MPI is an example of such a package. - /opt/hptc
Reserved for the exclusive use of the HP XC System Software. HP XC specific software and associated software packages are maintained under this directory. For more information, see “HP XC System Software Directory, /opt/hptc”. This directory also contains subdirectories for HP XC service configuration files. For more information on those directories, see “HP XC Service Configuration Files”. - /usr
Maintains a hierarchy of standard commands and files. - /usr/bin
Holds the binary executable files that any user can invoke. - /usr/sbin
Holds the binary executable files that are specifically for the system administrator or designated privileged users. - /usr/local
Provides an area to store executable programs that were produced locally. - /var
Holds various log, spooling, and temporary storage directories. - /tmp
Provides temporary file storage.
New third-party software packages may break the system. Be sure to assign individual areas when using relocatable software packages, that is, do not intermingle other vendors' software with the HP XC system files directories. Systemwide Directory, /hptc_clusterThe /hptc_cluster directory is the global file system on an HP XC system. This file system is shared and mounted by all nodes. This directory contains configuration and log file information that is applicable across the system; various services rely on the files in this directory. These log files are in the /hptc_cluster/adm/logs directory. Use the following guidelines for the /hptc_cluster directory: Keep this directory small. Ensure that the minimum amount of data is maintained in this directory. Ensure that the /hptc_cluster file system utilization is less than 90 percent. You can use the df command to determine the current value. Also, Nagios reports on high utilization based on thresholds in the nagios_vars.ini file; for more information, see Chapter 8 Minimize the number of nodes that access this storage. Do not install install software in this directory unless you are instructed to do so. Do not use the /hptc_cluster directory for parallel job storage and synchronization. Instead, create another file system and mount it internally. See Chapter 17 for more information. Do not use this directory for home directories or user accounts.
During system installation, you are given an option to locate the /hptc_cluster file system somewhere other than the installation on the head node. For example, you can place it on a partition on the head node's disk, on a networked disk, on an HP SFS server, or on a SAN. This directory is mounted on the head node. You must ensure the persistence of this file system mount. HP XC System Software Directory, /opt/hptcThe HP XC System Software maintains the /opt/hptc directory for its exclusive use. Its software is installed in that directory. Do not install or replace any other software in this directory unless it is an officially supported patch. Software packages, such as the SLURM resource manager, are installed in directories under the /opt/hptc directory under their own names. Software packages under the /opt/hptc directory have their own subdirectories, usually bin, etc, sbin and presumably others). There are symbolic links to files in these subdirectories (for example, /opt/hptc/package/bin) to the corresponding directory (for example, /opt/hptc/bin). The SLURM resource manager is such a package; its structure in the /opt/hptc directory structure is illustrated in Figure 1-2; the SLURM directory and its subdirectories are shown as dashed line boxes. The SLURM package has its own bin, etc, lib, and sbin, subdirectories. Each executable in the /opt/hptc/slurm/bin directory has a symbolic link to the /opt/hptc/bin directory. Likewise, there are symbolic links in the other SLURM subdirectories. HP XC Service Configuration FilesThe /opt/hptc/etc/ directory includes several subdirectories containing scripts used to configure services on nodes at installation time. The /opt/hptc/etc/sconfig.d directory contains scripts for system configuration. The /opt/hptc/etc/gconfig.d directory contains scripts used to gather information needed to configure a service on the HP XC system. The /opt/hptc/etc/nconfig.d directory contains scripts associated with specific services, which configure nodes that run these services . The /opt/hptc/etc/cconfig.d directory contains scripts associated with client-specific services. Log Files | |
The software components that comprise the HP XC system generate their own log files. As discussed in Chapter 7: Monitoring the System, some information is aggregated from local nodes to regional nodes and from regional nodes to the global node. Table 1-1 lists the components and the pathname of their log files. Table 1-1 Log Files | Component | pathname of Log File |
|---|
Console Management Facility (CMF) | /hptc_cluster/adm/logs/cmf.dated/* | collectl utility | /var/log/collectl | LSF | /opt/hptc/lsf/top/log (linked to /var/lsf/log) | Myrinet® gm_drain_test | /var/log/diag/myrinet/gm_drain_test/ | Myrinet gm_prodmode_mon diagnostic tool | /var/log/diag/myrinet/gm_prodmode_mon/links.log /hptc_cluster/adm/logs/aggregator_nodename.log (alerts) | ovp | /hptc_cluster/adm/logs/ovp/ovp_nodename_mmddyy[rnn] /hptc_cluster/adm/logs/ovp/current_ovp_log (a symbolic link to the most recent log file) | powerd | /var/log/powerd/powerd.log | Quadrics qselantestp | /var/log/diag/quadrics/qselantestp/ (default directory) | Quadrics qsnet2_drain_test utility | /var/log/diag/quadrics/qsnet2_drain_test | SLURM | /var/slurm/log/slurmctld.log and /var/slurm/log/slurmd.log | syslog-ng | /hptc_cluster/adm/logs/aggregator_nodename.log (one log file for each aggregator) /hptc_cluster/adm/logs/consolidated.log | startsys/stopsys | /hptc_cluster/adm/logs/startsys.log | systemimager/flamethrower | /hptc_cluster/adm/logs/imaging.log | cluster_config | /var/log/nconfig.log (contains output from the cluster_config utility) | Nagios | /opt/hptc/nagios_log /hptc_cluster/adm/logs/sel/sel-nxxx (System Event log) | transfer_to_avail/ transfer_from_avail | /var/log/transfer_avail.log | Fresh installation | /var/log/postinstall.log (shows the RPMs that were installed) | Software upgrade | /var/log/upgrade (if this directory exists, a software upgrade was done) |
|
Printable version
