Virtualization, High Performance Computing, Healthcare IT, Enterprise Computing

Articles by Rod

You are currently browsing Rod’s articles.

We had a customer using the free Microsoft SQL Server database with vCenter 5.0 that had surpassed the 4GB limit, at which point vCenter started crashing.

To fix the issue, stop the vCenter service, and connect to the vCenter database with SQL Server express. Make a backup of the database before you start.

Next issue the following commands:

truncate table VPX_HIST_STAT1;
truncate table VPX_SAMPLE_TIME1;

truncate table VPX_HIST_STAT2;
truncate table VPX_SAMPLE_TIME2;

truncate table VPX_HIST_STAT3;
truncate table VPX_SAMPLE_TIME3;

truncate table VPX_HIST_STAT4;
truncate table VPX_SAMPLE_TIME4;

Open the vCenter database dbo.VPX_PARAMETER. Adjust the following parameters in the table:

event.maxAge   30
event.maxAgeEnabled  true
task.maxAge 30
task.maxAgeEnabled true

After making these changes go to the programmability folder in Sql Server Management Studio, click stored procedures, execute dbo.cleanup_event_tasks_proc. This might take a while.

After that completes, right click on the databases and go to shrink database to get it under 4GB. You can restart vCenter now.


With the almost weekly Java exploits it is a best security practice to disable Java in your browser. FireFox on Linux will do this automatically. (See about:plugins to check this)

Some applications like WebEx still rely on Java on Linux. here are the ways to get the Oracle Java Plugin integration working after downloading the very latest Oracle Java JRE if you need to have it running:

#install latest RPM from Oracle
rpm –i jre-7u17-linux-x64.rpm
#Configure it as the default Java install
alternatives --install /usr/bin/java java /usr/java/latest/bin/java 20000
alternatives --install /usr/bin/javaws javaws /usr/java/latest/bin/javaws 20000
#Configure the Java plugin for 64-bit Firefox
alternatives --install /usr/lib64/mozilla/plugins/libjavaplugin.so libjavaplugin.so.x86_64 /usr/java/jre1.7.0_17/lib/amd64/libnpjp2.so 21000
#Check it is indeed being used
alternatives --display libjavaplugin.so.x86_64
#Test. Be sure that the version displayed is the version installed
firefox http://www.java.com/en/download/testjava.jsp

Tags: ,

Veeam has announced the next release of Veeam Reporter Enterprise, now known as Veeam Reporter 4.0. This release is currently in Beta.

This new release builds on Microsoft SQL Server Reporting Services (SSRS), which will benefit customers by removing the need to have Microsoft office applications installed for reports, allows access from any client, and allows Veeam to ship reporting packs faster and outside of major releases. Note there is no charge for SSRS with SQL Server Express (free.)

There is a new web dashboard for 4.0, this new dashboard can be customized and embedded into other websites like Sharepoint.

Veem Reporter 4.0 now includes historical performance data which allows for trouble shooting and capacity planning.

Finally new functionality has been added to support new vSphere technologies such as host profiles and distributed switches, integration with Veeam Business View, and Visio diagram support for up to 6000 VM’s.

Veeam Backup and Replication 5.0 now has TWO editions, Standard Edition 5.0, and Enterprise Edition 5.0

Note that ALL THE CURRENT FEATURES of Veeam Backup and Replication 4.1.1 are in Standard Edition, plus all the new 5.0 features. NO current features have been removed and moved to the Enterprise edition, which is very good for existing customers. All existing Veeam Backup and Replication Customers ( on maintenance) will get 5.0 Standard Edition at no charge.  Veeam is creating an Enterprise Edition to target new high end features that not all customers may want to pay a premium for.

Currently, there are differences in three areas between Standard and Enterprise Edition:

U-AIR (universal application-item recovery): Enterprise Edition includes U-AIR™;
Standard Edition does not. U-AIR is the ability to restore individual application objects,
such as email messages, directory objects or database records, from a regular imagelevel
backup. To the extent this capability is available in the market today (some vendors
offer very restrictive and limited implementations), it is considered a “premium” capability
and carries an additional cost. Veeam is excited to be able to offer a much more flexible
and robust implementation that brings greater value to customers.

Recovery verification: Standard Edition provides basic recovery verification (starting
a VM from a backup file and manually testing it), while Enterprise Edition offers
automated and more advanced recovery verification (including a fully managed,
isolated test environment; support for application test scripts; and bulk VM testing).

On-demand sandbox: Enterprise Edition includes the ability to run one or more VMs
from backup in an isolated environment, providing a working copy of the production
environment for troubleshooting, testing, training, etc. Standard Edition does not
include this capability.

Windows guest file system indexing: Standard Edition maintains an index of Windows
guest files currently on disk; Enterprise Edition also includes files in archived backups.

Customers who are current on maintenance will receive Standard Edition v5.0 as a regular product update at no additional charge. Standard Edition will include all the functionality in the current v4.1 software plus the major new features described in this blog, and the usual fixes and improvements.

For customers that want the enhanced functionality of Enterprise Edition,  the Enterprise upgrade fee will be waived on licenses purchased on or before June 18th, 2010. You must have maintenance on all sockets through the end of the year. ( We recommend customers always stay on maintenance, so this should be a non-issue).

Many customers want to know if they should upgrade to Enterprise, or stick with Standard?  For customers that want the enterprise features mentioned, you should upgrade for no cost. The caveat is that all FUTURE purchases of Enterprise will be $300 per socket MORE expensive than Standard edition.  MSRP is 599 on Standard, 899 on Enterprise. Also annual maintenace will increase from $119 a socket for standard to $179 a socket for Enterprise.

To upgrade your license to Enterprise, go to http://www.veeam.com/manage_licenses.html

If there are any questions on the new Veeam Backup and Replication 5.0, please contact HiperLogic and we will help you determine what is best for your particular situation

A customer requested to have a a quick and dirty script copied from d:\scripts to c:\temp and executed across a few hundred XP boxes ( not a cluster in this case ). Just in case you are looking to do this:

For %%a in ( 10.2.03) do start Robocopy d:\scripts\ \\%%a\C$\temp /e
psexec @hostlist c:\temp\doit.bat

I found that WinRM limits the number of concurrent shells while developing some PowerShell tools that connect around to a bunch servers in parallel. You may get an error like:

“Connecting to remote server failed with the following error message : The WS-Management service cannot process the request. This user is allowed a maximum number of 4 concurrent shells, which has been exceeded. Close existing shells or raise the quota for this user.”

This is how to configure and override the default of 5:

PSH>winrm quickconfig
PSH>winrm get winrm/config
PSH>winrm set winrm/config/winrs `@`{MaxShellsPerUser=`”50`”`}


ILO is a management port on HP servers that allow you to power nodes on and off, obtain sensor information, and other useful management tasks. You can save a lot of time when trying to configure the ILO management port on HP cluster nodes by scripting the ILO configuration with HP tools.

At a high level, to automatically configure ILO you feed the HP hponcfg.exe tool XML fragments which automatically configure ILO to your specifications.

I typically like the ILO management port to be configured to use DHCP, set the ILO hostname to be -ilo, and have SSH enabled. I normally set the username and password to be the same on each cluster blade.
First, download the hponcfg.exe tool from the HP website and put it on a share accessible to the compute nodes.

The first item it to set all compute nodes to have the same ILO name and password. In this example, I will just set them to Administrator/Administrator ( obviously you will want to change that!)

To reset the password create a file named reset_pass.xml on a share accessible from the compute nodes with the following contents:

 <LOGIN USER_LOGIN="Administrator" PASSWORD="Administrator">
  <USER_INFO MODE="write">
   <MOD_USER USER_LOGIN="Administrator">
    <PASSWORD value="Administrator"/>

Then run this command to execute the password reset on all nodes, via clusrun or other mechanism ( psexec if you are not on HPC Server 2008):

hponcfg /f reset_pass.xml

Next, lets reset the ILO hostname to -ilo and enable DHCP. Create a batch file with the following contents, copy it to the temp directory on each node, and execute it on each node via clusrun.

for /f "delims=" %%a in ('hostname') do @set HOSTNAME=%%a
echo ^<RIBCL VERSION="2.1"^> >> network.xml > network.xml
echo ^<LOGIN USER_LOGIN="Administrator" PASSWORD="Administrator"^> >> network.xml
echo   ^<DIR_INFO MODE="write"^> >> network.xml
echo   ^<MOD_DIR_CONFIG^> >> network.xml
echo     ^<DIR_AUTHENTICATION_ENABLED VALUE = "N"/^> >> network.xml
echo     ^<DIR_LOCAL_USER_ACCT VALUE = "Y"/^> >> network.xml
echo     ^<DIR_SERVER_ADDRESS VALUE = ""/^> >> network.xml
echo     ^<DIR_SERVER_PORT VALUE = "636"/^> >> network.xml
echo     ^<DIR_OBJECT_DN VALUE = ""/^> >> network.xml
echo     ^<DIR_OBJECT_PASSWORD VALUE = ""/^> >> network.xml
echo     ^<DIR_USER_CONTEXT_1 VALUE = ""/^> >> network.xml
echo     ^<DIR_USER_CONTEXT_2 VALUE = ""/^> >> network.xml
echo     ^<DIR_USER_CONTEXT_3 VALUE = ""/^> >> network.xml
echo   ^</MOD_DIR_CONFIG^> >> network.xml
echo   ^</DIR_INFO^> >> network.xml
echo   ^<RIB_INFO MODE="write"^> >> network.xml
echo   ^<MOD_NETWORK_SETTINGS^> >> network.xml
echo     ^<SPEED_AUTOSELECT VALUE = "Y"/^> >> network.xml
echo     ^<NIC_SPEED VALUE = "10"/^> >> network.xml
echo     ^<FULL_DUPLEX VALUE = "N"/^> >> network.xml
echo     ^<DHCP_ENABLE VALUE = "Y"/^> >> network.xml
echo     ^<DHCP_GATEWAY VALUE = "Y"/^> >> network.xml
echo     ^<DHCP_DNS_SERVER VALUE = "Y"/^> >> network.xml
echo     ^<DHCP_STATIC_ROUTE VALUE = "Y"/^> >> network.xml
echo     ^<DHCP_WINS_SERVER VALUE = "Y"/^> >> network.xml
echo     ^<REG_WINS_SERVER VALUE = "Y"/^> >> network.xml
echo     ^<DNS_NAME VALUE = "%HOSTNAME%"/^> >> network.xml
echo     ^<DOMAIN_NAME VALUE = "mydomainname.com"/^> >> network.xml
echo     ^<SEC_WINS_SERVER value = ""/^> >> network.xml
echo     ^<STATIC_ROUTE_1 DEST = "" GATEWAY = ""/^> >> network.xml
echo     ^<STATIC_ROUTE_2 DEST = "" GATEWAY = ""/^> >> network.xml
echo     ^<STATIC_ROUTE_3 DEST = "" GATEWAY = ""/^> >> network.xml
echo   ^</MOD_NETWORK_SETTINGS^> >> network.xml
echo   ^</RIB_INFO^> >> network.xml
echo  ^</LOGIN^> >> network.xml
echo ^</RIBCL^> >> network.xml

hponcfg /f network.xml

Finally, lets enable SSH for ILO. Put the following contents in a file called ssh.xml, copy it to a share and run hponcfg /f ssh.xml via clusrun

  <LOGIN USER_LOGIN="Administrator" PASSWORD="Password">
  <RIB_INFO MODE="write">
      <SSH_PORT value="22"/>
      <SSH_STATUS value="Yes"/>

For more information, read the extensive HP documentation and examples.


Maya is a popular network renderer used for Hollywood movies and a variety of other applications. It can be used on clusters of workstations, as well as HPC Server 2008. While there is no native HPC Server 2008 integration as of this post, it is fairly straight-forward to integrate Maya’s mental ray renderer with some scripting.

The scripting language I chose was Perl,  though any language you like will work. ( You can download Perl for Windows free from ActiveState).

Before you begin do the following:

1. set


on each compute node to the license server for Maya.
2. set the PATH to the Maya Mental Ray renderer, which by default is

c:\Program Files\Autodesk\mrstand3.7.1-maya2009\

3. Make sure the designers export to the “.mi” format which is the Mental Ray format.
4. Finally, make sure the designers are referencing resources using relative paths, and not hardcoded paths like c:\myresoure.img which will not exist at render time on the compute nodes.

The custom script I created called “render.pl” takes a few arguments , the job ( In .mi format, required for Mental Ray rendering), the start frame, the end frame, and the number of frames per job.

   render.pl -i "job.mi"  -s 1 -e 800 -f 4

Each job (or task) takes a chunk of the frames and renders them to images/, the more nodes you have the more you can render in parallel to speed up the render job.

The script generates the HPC Server 2008 jobs  based on the user input, ultimately calling the Maya Mental Ray renderer with

 mentalrayrender.cmd -threads 8 -render $start $end $input

The advantage of using HPC Server 2008 for Maya is HPC Server 2008 has all the built in scheduling, job management, and job submission built in. In this case the cluster was being used for many other applications, so Maya was able to play nicely with the existing cluster.


A common management task is to set an environment variable of some sort across a compute cluster.

To set an environment across the cluster, you can use cluscfg:

cluscfg setenvs name=value

For example

cluscfg setenvs LM_LICENSE_FILE=license_server@4000

Note this will apply only to batch jobs started with the Microsoft HPC scheduler, not to normal programs running outside of batch.

If you need to set a variable across all nodes, and need it to be referenced outside of user batch jobs, then do the following:

Select the nodes into the node management view and run the following command 

setx  name value /M

This command will add this environement variable in the system of all the selected nodes:

clusrun setx PATH  "%PATH%;\\headnode\soft\bin"


© 2006-2010 HiperLogic, LLC.  |  Serving the Ann Arbor, Southeast Michigan, and Ohio region  |  (888)-268-3930  |