Windows HPC Server 2008 Service Pack 2 Release Notes
Applies To: Windows HPC Server 2008
These release notes address late-breaking issues and information about this release of Windows® HPC Server 2008 Service Pack 2. These notes will help you to install and use this product.
In these release notes:
Installing Windows HPC Server 2008 Service Pack 2
Known issues
Installing Windows HPC Server 2008 Service Pack 2
Be aware of the following items and recommendations before you start installing Windows HPC Server 2008 Service Pack 2 to upgrade an existing cluster from Windows HPC Server 2008 Service Pack 1:
To install Windows HPC Server 2008 Service Pack 2, you must have Windows HPC Server 2008 Service Pack 1 already installed.
Create backups of the HPCClusterService and the HPCReporting databases before installing Service Pack 2. When you install the service pack, new indexes and new parameters for some procedures are added to the HPC databases. Uninstalling Service Pack 2 is not possible. If the installation fails, you must completely uninstall HPC Pack 2008, install it again, install Service Pack 1, restore the databases from a backup, and then follow the additional steps described in Steps to Perform Before and After Restoring the HPC Databases from a Backup.
Close all open windows for applications related to the Microsoft® HPC Pack 2008 before you apply the service pack. If you have applications open, such as HPC Cluster Manager, you may be prompted to restart your computer to complete the installation.
Do not apply the service pack during critical times or while a long running job is still running. When you upgrade either a head node or a compute node in a cluster, you may be prompted to reboot the computer to complete the installation.
When you apply the service pack to your head node, the files that are used to deploy a compute node are also updated. If you install a new compute node or redeploy an existing compute node, the service pack is automatically applied to that compute node.
You can also use the Microsoft Update management feature of the HPC Pack 2008 to apply the service pack to existing compute nodes. To use the Microsoft Update management feature:
View nodes by node template to identify the node templates for the compute nodes that you want to update. For information about how to view nodes by node template, see “View Nodes” in HPC Cluster Manager Help in the Windows HPC Server 2008 Technical Library (https://go.microsoft.com/fwlink/?LinkId=124146).
Add the Apply Updates task to the node templates for the nodes that you want to update, and include HPC Pack 2008 Service Pack 2 in the list of updates that you specify in the Updates parameter. For information about how to add the Apply Updates task to a node template, see “Add the Apply Updates Task to a Node Template” in HPC Cluster Manager Help in the Windows HPC Server 2008 Technical Library (https://go.microsoft.com/fwlink/?LinkId=124146).
Run the Maintain node action on the compute nodes. For information about how to run the Maintain action on a compute node, see “Run Maintenance Tasks on Compute Nodes” in HPC Cluster Manager Help in the Windows HPC Server 2008 Technical Library (https://go.microsoft.com/fwlink/?LinkId=124146).
Verify that the updates were applied by checking the software version number on the compute nodes. To view the software version number for your compute nodes, you can add the Version column to the node List View in HPC Cluster Manager. The updated version number starts with 2.2 (for example, 2.2.1841.0).
If the service pack was not applied, verify that your cluster’s updates are coming from a WSUS server and that the HPC Pack 2008 Service Pack 2 update has been approved by your WSUS administrator.
In the Windows Update control panel, if you see that You receive updates: Managed by your system administrator does not have a link associated with it, then you are currently getting updates from a WSUS server.
Click Check online for updates from Windows Update to see if the service pack appears. If it does, then contact your WSUS administrator and ask them to approve the HPC Pack 2008 Service Pack 2 update.
Note that because the service pack stops the HPC services during installation, the provisioning log will not contain information about the service pack being installed.
If you have set up a head node that has failover clustering configured, use the following steps to apply the service pack:
Apply the service pack to the active node.
A failover will occur as the node reboots. Ensure that the failover cluster is reporting that the head node is healthy and online.
Ensure that the head node is now running on the other server in the failover cluster, and then apply the service pack to that server.
Known issues
The known issues for Windows HPC Server 2008 also apply to Windows HPC Server 2008 Service Pack 2. For information about the known issues for Windows HPC Server 2008, see Windows HPC Server 2008 Release Notes.
The following are the known issues specific to Windows HPC Server 2008 Service Pack 2:
HPC Cluster Manager may not display the correct number of cores for a compute node
Manually uninstalling software updates can give inconsistent diagnostic results
Deployment of compute nodes on virtual machines might fail when using localized operating system versions
The mpicsync tool might fail when analyzing the trace files for an MPI program that failed
Unreachable head nodes cannot be removed from the node list
HPC Cluster Manager may not display the correct number of cores for a compute node
HPC Cluster Manager may not display the correct number of cores for a compute node if that compute node was originally deployed with hyper-threading enabled, and hyper-threading was disabled after deployment.
Workaround
In HPC Cluster Manager, take the node offline, and then bring the node online again. The correct number of cores will be displayed after the compute node is back online.
Known issues
Manually uninstalling software updates can give inconsistent diagnostic results
If you manually uninstall software updates and then run the Installed Software Updates Report diagnostic test, the software updates that you uninstalled will still be listed in the test results, as part of the software updates that are still installed. However, the software updates that you uninstalled will be correctly listed in the Pending Software Updates and Software Updates Required diagnostic test results. Additionally, if you run the Maintain action on a node, the software updates that you uninstalled will be correctly installed again.
Known issues
Deployment of compute nodes on virtual machines might fail when using localized operating system versions
For automated deployment to work correctly, the time zone of the head node must match the time zone of the compute nodes. When a localized operating system is installed on a virtual machine that will serve as a compute node, the operating system on the virtual machine will not default to the correct time zone initially, causing the deployment to fail in some instances.
Workaround
Temporarily change the time zone on the head node to match the time zone of the virtualization server. Alternatively, install the operating system and HPC Pack 2008 manually on the compute nodes.
Known issues
The mpicsync tool might fail when analyzing the trace files for an MPI program that failed
When analyzing the trace files of a Message Passing Interface (MPI) program, the clock synchronization tool (mpicsync) might fail to update the event time-stamps for the trace files. This can occur if the MPI program failed while it was actively using MPI communication calls, and was running on three or more nodes.
Known issues
Unreachable head nodes cannot be removed from the node list
On an HPC cluster that has the head node configured in a failover cluster, if one of the head node servers permanently fails and has to be removed from the cluster, the node entry for that head node server in the node list will remain as Unreachable.
Workaround
Use the Remove-HpcNode cmdlet in HPC PowerShell to remove the head node server that failed. For more information, see Remove-HpcNode.
Known issues