Unable to deploy PyTorch VM instance

Hi, I am trying to deploy PyTorch VM instance, after filling in the information, as per the online documentation in Google Cloud, I am running into errors, I am not sure what is causing it.

 

Screenshot 2024-04-25 at 1.22.38 PM.png

Solved Solved
1 7 138
2 ACCEPTED SOLUTIONS

@digs1998 

You are trying to deploy this Pytorch VM under the project, which has organization policy constraint preventing External IP for VMs. It means, that VMs under your project or most probably under entire organization are not allowed to get assigned External IP. 

DamianS_0-1714230188286.png

If you have Organization Policy permissions, add your project as an exception , by adding tag to project and into constraint, or disable vmExternalIpAccess organization policy.

DamianS_1-1714230371284.png

 


--
cheers,
DamianS
LinkedIn medium.com Cloudskillsboost

View solution in original post

Or you during deployment set ExternalIP to none

DamianS_2-1714230490913.png

But then you must configure either Cloud NAT ( for access to the Internet ) and IAP for being able to log in via ssh 
https://cloud.google.com/nat/docs/gce-example

 

View solution in original post

7 REPLIES 7

Hello @digs1998  ,Welcome on Google Cloud Community.

Are you able to provide link for mentioned documentation? It looks like there is an issue either with provided information by user or directly under the deployment source code. 

Update: I've followed this tutorial https://cloud.google.com/deep-learning-vm/docs/pytorch_start_instance and instance has been deployed without any issues:

DamianS_0-1714120141052.png


--
cheers,
DamianS
LinkedIn medium.com Cloudskillsboost

Hi @DamianS, thank you for your response, I have been following this tutorial itself, yet I am getting the errors I mentioned above.

Hi @DamianS Let me show you my approach while creating the VM instance here based on the template, it seems there is no Pytorch 1.8 + fast ai option, so I chose Pytorch 2.0 with CUDA 11.0.

Screenshot 2024-04-26 at 10.11.29 AM.png

Screenshot 2024-04-26 at 10.11.58 AM.png

I am still getting the error below.

Screenshot 2024-04-26 at 10.12.41 AM.png

@digs1998 

You are trying to deploy this Pytorch VM under the project, which has organization policy constraint preventing External IP for VMs. It means, that VMs under your project or most probably under entire organization are not allowed to get assigned External IP. 

DamianS_0-1714230188286.png

If you have Organization Policy permissions, add your project as an exception , by adding tag to project and into constraint, or disable vmExternalIpAccess organization policy.

DamianS_1-1714230371284.png

 


--
cheers,
DamianS
LinkedIn medium.com Cloudskillsboost

Or you during deployment set ExternalIP to none

DamianS_2-1714230490913.png

But then you must configure either Cloud NAT ( for access to the Internet ) and IAP for being able to log in via ssh 
https://cloud.google.com/nat/docs/gce-example

 

Hi @DamianS thank you for the suggestions, this does help, it seems I did not have some permissions earlier, I was able to deploy the instances now.

Wonderful. Happy to help 😉

--
cheers,
DamianS
LinkedIn medium.com Cloudskillsboost