Azure CycleCloud Slurm template VMs - some OS's don't startup sucessfully

Gary Mansell 111

Hi,

If I use the CentOS 7 OS choice for HPC VMs then they start OK, but I have also tried AlmaLinux and SLES15 HPC - and neither of these work.

The error with both is (and I have left it waiting for some time now). Otherwise, most of the other settings are per defaults:

I would like to at least be able to try using a more recent OS image for performance issues I am seeing - how do I get this to work?

Thanks

Gary

Ravi Kanth Koppala 3,231 Reputation points Microsoft Employee

2022-12-18T07:02:47.903+00:00
@Gary Mansell ,

It looks like there is a problem with the network configuration on the virtual machine (VM) that is preventing it from starting up successfully. The error message indicates that the VM is unable to execute the nslookup command to look up the IP address of the host riclurm-hpc-pq0-1.

There are a few potential causes for this issue:

Network connectivity issues: Ensure the VM has a valid IP address and can reach the internet or other required resources.

DNS issues: If the VM cannot resolve hostnames, this could prevent it from starting up properly. Ensure that the VM is configured to use a valid DNS server and that the server is reachable.

Incorrect configuration: Ensure the VM is properly configured to use the correct network settings, including the correct hostname and IP address.

Hardware or software issues: If the problem persists after checking the above items, a hardware or software issue may prevent the VM from starting. In this case, you may need to troubleshoot further to identify the root cause of the problem.

I hope this helps!
Gary Mansell 111 Reputation points

2022-12-19T10:31:45.42+00:00

I am just using the default AzureCycle Cloud 8.3 vm images, with no special configuration.

The Centos 7 & Ubuntu images work fine when I select them as the OS for the HPC nodes, so the Azure VM & networking config must be OK, and hence this must be an issue with the SLES* and AlmaLinux images or the CycleCloud Slurm template (surely)?
Ravi Kanth Koppala 3,231 Reputation points Microsoft Employee

2022-12-19T17:00:30.187+00:00
@Gary Mansell ,

There may be an issue with the SLES or AlmaLinux images or with the CycleCloud Slurm template. Without more information, it is difficult to determine the specific cause of the issue you are experiencing.

There are a few steps you can take to try to troubleshoot the problem:

Check the logs for any error messages or clues about what might be causing the issue.

Check the configuration of the SLES or AlmaLinux images and compare it to the configuration of the CentOS or Ubuntu images to see if there are any differences that might be causing the issue.

Check the configuration of the CycleCloud Slurm template and compare it to a working template to see if there are any differences that might be causing the issue.

If you are using a custom image or template, try using a default image or template to see if the issue persists.

If you are still unable to resolve the issue, you may want to reach out to Microsoft support or the community for further assistance.

1 answer

Gary Mansell 111 Reputation points

2022-12-22T17:51:51.213+00:00

The reason the AlmaLinux8 HPC nodes would not initialise was because the scheduler node was CentOS7 - it seems that you have to use AlmaLinux8 on both the scheduler and the HPC nodes
Please sign in to rate this answer.

0 comments No comments
Sign in to comment

Share via

Azure CycleCloud Slurm template VMs - some OS's don't startup sucessfully

1 answer