Difference between revisions of "Bright:Intel Cluster Ready"

From Define Wiki
Jump to navigation Jump to search
(Created page with "* Assumes that all the pacakges were installed as part of the default installation (Refer to the admin guide to install if not) == Prepare the system == Need to run as <tt>cm...")
 
 
(One intermediate revision by the same user not shown)
Line 20: Line 20:
  
 
== Handling Test Failures ==
 
== Handling Test Failures ==
The cluster checker produces several output files, with .xml, .out, .debug suffixes, which include time stamps in the filenames. If tests fail, the out- put files can be consulted for details. The output files can be found in the  ̃/intel-cluster-ready/logs directory.
+
The cluster checker produces several output files, with .xml, .out, .debug suffixes, which include time stamps in the filenames. If tests fail, the out- put files can be consulted for details. The output files can be found in the  ̃<tt>/intel-cluster-ready/logs<tt> directory.
When debugging and re-running tests, the option <tt>---include_only test_name </tt>
+
 
can be passed to cluster-check to execute only the test named “<test>” (and the tests on which it depends).
+
 
 +
When debugging and re-running tests, the option <tt>---include_only test_name </tt> can be passed to cluster-check to execute only the test named “<test>” (and the tests on which it depends).
 +
 
 +
 
 
In a heterogeneous cluster the cluster check run fails as a result of hardware differences. To resolve the failures, it is necessary to create mul- tiple groups of homogeneous hardware. For more information, the Intel Cluster Checker documentation can be consulted.
 
In a heterogeneous cluster the cluster check run fails as a result of hardware differences. To resolve the failures, it is necessary to create mul- tiple groups of homogeneous hardware. For more information, the Intel Cluster Checker documentation can be consulted.

Latest revision as of 16:59, 16 January 2014

  • Assumes that all the pacakges were installed as part of the default installation (Refer to the admin guide to install if not)

Prepare the system

Need to run as cmsupport

[root@mycluster ~]# su - cmsupport
[cmsupport@mycluster ~]$ clck-prepare
Created non InfiniBand node list file /home/cmsupport/intel-cluster-ready/nodelist
Created InfiniBand node list file /home/cmsupport/intel-cluster-ready/nodelist.ib

Run Cluster Checker

[root@mycluster ~]# su - cmsupport
[cmsupport@mycluster ~]$ module initadd intel-cluster-runtime
[cmsupport@mycluster ~]$ module load intel-cluster-runtime
[cmsupport@mycluster ~]$ cluster-check --certification

Handling Test Failures

The cluster checker produces several output files, with .xml, .out, .debug suffixes, which include time stamps in the filenames. If tests fail, the out- put files can be consulted for details. The output files can be found in the ̃/intel-cluster-ready/logs directory.


When debugging and re-running tests, the option ---include_only test_name can be passed to cluster-check to execute only the test named “<test>” (and the tests on which it depends).


In a heterogeneous cluster the cluster check run fails as a result of hardware differences. To resolve the failures, it is necessary to create mul- tiple groups of homogeneous hardware. For more information, the Intel Cluster Checker documentation can be consulted.