Get-HpcNodeState unrecognized in HeadNode

Question

I have deployed the Sample HPC Azure project to the Cloud. The deployment has 1 head node, 1 front end node and 98 compute nodes.

Although everything looks just fine in Azure Management Portal, there are a few problems:

1. On deployment, 17 nodes are unreachable.
2. Running Cluster manager on the head node fails with:

   Failed to communicate with remote SDM store. Connection Failed. 
   No connection could be made because the target machine actively 
   refused it ip.of.my.srvr:9893
   

3. No HpCNode cmdlet is available on the head node.
4. Starting a Job on unreachable node stays queued saying the requested resources are not available.

There is almost noway that I could manage the Nodes.

What is it that I'm missing?

Answer 1

Hi,

I am trying to involve someone familiar with this topic to further look at this issue. There might be some time delay.

Appreciate your patience.

Best Regards,

Ming Xu.

---

Please mark the replies as answers if they help or unmark if not.
If you have any feedback about my replies, please contact [e-mail].
Microsoft One Code Framework

Answer 2

Do you try sample at http://msdn.microsoft.com/en-us/library/hh560251(v=vs.85).aspx and install the SDK?

Do you try to reduce the instance to 10 and try again?

Answer 3

I did install the sample. When the size is 10, I don't see any unreachable nodes. but when I go to a larger cluster (which I need, and I have 300 available nodes), I see about 10%-20% unreachable nodes.

Anyhow, problems 2 and 3 still exist on a small HPC cluster.

Answer 4

Thank you very much. with this problem, I'm not using 10%-20% of my 300 nodes. At first (when I didn't have this many nodes), I assumed it's by design (to reserver a few nodes), but when I looked closer, I found out they are unreachable.

What is interesting for me was the inability to use Any HpCNode cmdlets such as:

• Restart-HpcNode, Remove-HpcNode
• Get-HpcNode, Get-HpcNodeStateHistory,...
• Set-HpcNode, Set-HpcNodeState

Actually I'll go ahead and list all available *Hpc* cmdlets as a reply to the main question.

Answer 5

Here are the available cmdlets in the HeadNode:

Add-HpcPool Add-HpcTask Copy-HpcJobTemplate Export-HpcJob Export-HpcJobTemplate Export-HpcSoaSessionTrace Export-HpcTask Get-HpcClusterOverview Get-HpcClusterProperty Get-HpcJob Get-HpcJobCredential

Get-HpcJobTemplate Get-HpcJobTemplateAcl Get-HpcPool Get-HpcTask Import-HpcJobTemplate New-HpcJob Remove-HpcJobCredential Remove-HpcJobTemplate Remove-HpcPool Remove-HpcSoaCredential

Remove-HpcSoaSessionTrace Set-HpcClusterProperty Set-HpcJob Set-HpcJobCredential Set-HpcJobTemplateAcl Set-HpcPool Set-HpcSoaCredential Set-HpcTask Stop-HpcJob Stop-HpcTask Submit-HpcJob

Answer 6

I found you raise the SR and  confirmed all nodes are working now but  second question is unanswered, why HPC cmdlets are not available on the headnode.

• Restart-HpcNode, Remove-HpcNode
• Get-HpcNode, Get-HpcNodeStateHistory,...
• Set-HpcNode, Set-HpcNodeState

What is the detail error information when you run the Restart-HpcNode?

Do you have right permission for the current user context?

If you run the command from remote client, please check document below

http://blogs.technet.com/b/windowshpc/archive/2009/01/07/a-powershell-problem.aspx