This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(35.2, 85%, 13%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EBP%3C/text%3E%3C/svg%3E)
Hello,
in the modules / exercises:
a setup.sh script in the corresponding subdirectories of azure-ml-labs/Labs from the cloned Git repository https://github.com/MicrosoftLearning/mslearn-azure-ml.git should be executed. However, I get an error message stating that I do not have sufficient quota to create the compute cluster with Standard DSv2 Family Dedicated vCPUs (see screenshot). I am using the free Azure 200$ test subscription and cannot increase any quota limits (e.g. Standard DSv2 Family Cluster Dedicated vCPUs, there is no entry Standard DSv2 Family Dedicated vCPUs), as far as I can see.
Does the setup.sh script need to be modified or is there a workaround (I assume that many other people have already experienced the same issue)? I think doing the course exercises should be possible even with the free Azure test subscription.
Any help is appreciated.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(192, 43%, 28%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ER%3C/text%3E%3C/svg%3E)
Hi Proxauf, Bastian,
Thanks for reaching out to us on Microsoft Q&A forum.
We are looking into the issue will get back to you soon.
Hello Rdash,
thanks in advance. While exercise https://microsoftlearning.github.io/mslearn-azure-ml/Instructions/09-Run-pipelines.html yields the same error, https://microsoftlearning.github.io/mslearn-azure-ml/Instructions/09-Hyperparameter-tuning.html apparently does not (all resources were created successfully there). However as far as I can see, both exercises use the same setup.sh script inside azure-ml-labs/Labs/09.
Maybe the terminal output helps in clarifying the issue:
terminal_output_notworking.txt
Apart from the different region ('westus' and 'westeurope' in the one-attempt error and success cases, respectively), which appears to be randomly selected from ("eastus" "westus" "centralus" "northeurope" "westeurope") and random ID strings, I do not see any other significant differences, though.
I manually tested the five regions by modifying the setup.sh script. Apparently "westus" and "centralus" generate the error message, whereas "eastus", "westeurope" and "northeurope" do not.
This is also confirmed by the quotas for the different regions, where the non-working and working regions each show 0 and 6 cores available for Standard DSv2 Family Cluster Dedicated vCPUs, respectively.
Given the test subscription, as mentioned in the original post, I cannot change these quotas. A workaround is to hardcode the region variable to one of the above three working values.
May I propose to set the quota for test subscriptions in general to the same non-zero value of 6 for all five regions used by setup.sh or alternatively to modify the setup.sh script to automatically choose a region where sufficient quota is available?
This comment has been deleted due to a violation of our Code of Conduct. The comment was manually reported or identified through automated detection before action was taken. Please refer to our Code of Conduct for more information.
Hi Proxauf, Bastian,
Thank you for reaching out to us on the Microsoft Q&A forum.
It appears that the limit of vCores for the region has been surpassed.
To verify the current usage of vCPUs from the Azure portal:
Navigate to your subscription.
Access the "Usage + Quotes" section.
Locate the usage details for "Standard FS family" vCPUs.
To initiate a quota, increase request, proceed to the portal and submit a support ticket. Within the ticket, specify the region and request an augmentation of your quota.
Reference documents: https://learn.microsoft.com/en-us/azure/azure-resource-manager/troubleshooting/error-resource-quota?tabs=azure-cli#solution
https://learn.microsoft.com/en-us/azure/quotas/per-vm-quota-requests
Please don't hesitate to reach out to us if you have any further queries. I hope the information provided has been helpful to you! If so, please accept the answer by clicking the Accept Answer / Upvote on the post. We value your feedback, and it will help to assist others who might have a similar query. Thank you for your contribution in enhancing Microsoft Q&A!
If you still persist the issue, please feel free to reach out to us we will be happy to help you out.
Thank you.
Hi Rdash,
as mentioned, having a test subscription, I cannot increase the quota. Also I'm wondering why your refer to Standard FS family instead of Standard DSv2 Family Cluster Dedicated vCPUs.
Maybe I'm wrong, but to me it looks like this is an issue of the quotas from Azure AI (rather than of the overall subscription quotas accessible as described in your answer). For example, in a region that works (West Europe), these quotas show two used cores, namely the Standard DSv2 Family Cluster Dedicated vCPUs ones, and four remaining available cores:
The regions that do not work ("Central US", "West US") on the other hand do not have any cores available:
On a somewhat related note, in exercise https://microsoftlearning.github.io/mslearn-azure-ml/Instructions/11-Deploy-online-endpoint.html, the notebook azure-ml-labs/Labs/11/Deploy to online endpoint.ipynb by default uses one instance of Standard_F4s_v2 SKUs for the blue deployment. Also here the quota for the region (North Europe in this case) is too low: 6 cores are available, 8 cores would be needed (https://learn.microsoft.com/en-us/azure/virtual-machines/fsv2-series lists those as having 4 vCPUs; maybe those are dual-cores).
The only way to avoid that issue at the moment is to size down the machines, e.g. to Standard_F2s_v2, so that only 4 cores are being used:
Hi Proxauf, Bastian,
Since you mentioned that you're utilizing the free Azure $200 test subscription and are unable to request a quota on the Azure free trial account for Azure AI Studio and ML Labs, the solution would be to upgrade to a pay-as-you-go plan.
As shown in the screenshot:
If you still persist any issue, please feel free to reach out to us we will be happy to assist you further.
Upgrading to pay-as-you-go certainly is one possibility. However, as mentioned, it would be reasonable to have sufficient quota for a certificate course even with the free subscription. I don't think these courses require excessive computational power given that the virtual machines used are basically the default suggestions shown in Azure ML when creating compute. The practice labs included in the course would be a good showcase of what can be done with Azure Machine Learning, provided there were sufficient quota. In any case, as for now, I think, we can consider the topic closed.
If you require extra quotas beyond what the free subscription provides, you'll need to upgrade to a pay-as-you-go subscription. Additionally, please don't hesitate to share your suggestions and concerns with the Microsoft team.