Task 3: Run the OVP to Verify Software and Hardware Components

The operation verification program (OVP) verifies the major HP XC software and hardware components to provide a level of confidence that the system has been installed and configured correctly.

The OVP performs various tests to verify the following:

The interconnect is functional.
Network connectivity has been established.
The administration network is operational.
A valid license key file is installed and the license manager servers are up.
All compute nodes are responding and are available to run applications.
SLURM control daemons are responding and partitioning is valid if LSF-HPC with SLURM is configured.
LSF is up and running, if LSF is configured.
An MPI job can be compiled, launched, and executed (through the xring test)
Serial and parallel applications can be submitted and executed through LSF from all compute nodes, if LSF is configured.
CPU usage on all nodes except the head node (by default).
Memory usage on all compute nodes but not the head node (by default).

The OVP also runs the following benchmark tests. These tests compare values relative to each node and report results with values more than three standard deviations from the mean:

LINPACK is a collection of Fortran subroutines that analyze and solve linear equations and linear least-squares problems. This test is CPU intensive and stresses the nodes, with limited data exchange.
PALLAS exercises the interconnect connection between compute nodes to evaluate MPI performance. It provides a concise set of benchmarks targeted at measuring the most important MPI functions. It tests multiple nodes at one time.
STREAMS tests memory bandwidth.
The Network Unidirectional test runs ping pong ring to test the interconnect between two nodes at a time.

The documentation that was delivered with your model of HP cluster platform and the HP XC System Software Administration Guide describe interconnect diagnostic tests.

Start the Operation Verification Program

Follow this procedure to start the OVP:

Begin this procedure as the root user on the head node.

Start the verification procedure without any component-specific options to test the entire system:

# ovp [--verbose [--verbose]] [--timeout=0]




	NOTE: HP recommends that you use the single --verbose option to see details about tests as they are being run. Specify the double verbose option ( --verbose --verbose) to obtain more details about benchmarking test results. Specify the --timeout=0 option on systems with more than 128 nodes to turn off the time out limit of some tests.

Follow along with the OVP command output. Because the OVP command output spans several pages, it is shown in Appendix K. Return here when OVP processing is complete.
Examine the test results to ensure that all tests passed. Test results are stored in a date-stamped log file located in the /hptc_cluster/adm/logs/ovp directory.
Test failures and warnings are clearly reported in the log file, and it contains some troubleshooting information. In some cases, the errors might be obvious, and the test output is terse.
OVP Log File Naming Convention
The format of the OVP log file name includes the following:
- The internal name of the head node.
- The OVP run date in the format MMDDYY (2-digit month, 2-digit day, and 2-digit year).
- The run number, which represents the number of times the OVP has been run. The run number is not included for the first run.
For example, the log file name ovp_ n16_070607.log indicates that this is the first run of the OVP on 06 July 2007, on head node n16.
See “Troubleshooting the OVP” for more information about troubleshooting failed test results.
For information about verifying individual cluster components on demand, see ovp(8) and the HP XC System Software Administration Guide.
When all OVP tests pass, proceed to “Task 4: Run the SVA OVP Utility” (if SVA is installed) or “Task 5: View System Health”.

Task 3: Run the OVP to Verify Software and Hardware Components

Technical documentation

» Table of Contents

» Glossary

» Index

Start the Operation Verification Program

OVP Log File Naming Convention