Availability of GPU quota

[Michael IV]

Hi, I am trying to understand something in respect to launching an instance with GPU attached.

I requested a quota from google  for 1 K80 GPU in europe-west1 zone. I got it. I can see in the quota list   that I have "0./1"  usage status.

Now when I am trying to launch my instance to which I attach k80 GPU it fails with 

The zone 'projects/xxxxxxxxx/zones/europe-west1-b' does not have enough resources available to fulfill the request. Try a different zone, or try again later.

and I also see that while the instance is trying to start, in the Quota pages I can see that the GPU quota gets " 1/1" ,usage. But in reality

I can't gave unused GPU for the instance.

Why is that and how do I solve that? 

[Karthick (Cloud Platform Support)]

Hello Michael,

In order to better understand and verify, can you send me the exact parameters you are using to create the instance in europe-west1-b, so that I will try to reproduce and let you know?  Please send this in a private message.

p.s. Please note that I have concealed your project ID from the previous message as this is a sensitive information (Project ID).

[Paul Nash]

Hi Michael, just to build on the question from Karthick,

The "not enough resources" class of error is a response to your *entire* request. It is likely that you have tried to request other aspects of the VM that may not be available on machines that also have GPUs installed. We don't document the exact configurations because it's very complicated, but we're working on providing some hints in the future that can help you understand what is available and what is not. If you can answer Karthick's question and describe the instance that you're trying to create, we can likely tell you why it's not able to be created.

One final note to clarify: quota is not a promise that your request to create resources at any given time will succeed. It's a limit to how much you're able to ask for, but most quotas are set well above the limits of what a customer is likely to ask for given their typical usage.

Thx,

-P

--
© 2018 Google Inc. 1600 Amphitheatre Parkway, Mountain View, CA 94043
 
Email preferences: You received this email because you signed up for the Google Compute Engine Discussion Google Group (gce-dis...@googlegroups.com) to participate in discussions with other members of the Google Compute Engine community and the Google Compute Engine Team.
---
You received this message because you are subscribed to the Google Groups "gce-discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gce-discussio...@googlegroups.com.
To post to this group, send email to gce-dis...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gce-discussion/762c5413-a78e-4c1f-b6f0-7d0130cd1b8c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Paul R. Nash |

Group Product Manager, Compute Engine |

paul...@google.com |

206-876-1620

[Michael IV]

Hi.Thanks for your input. Here is the link to ServerFault where I put all the info regarding my issue,including the hardware config I am trying to use:

https://serverfault.com/questions/908648/google-compute-engine-unable-to-attach-gpu-to-instance

It is really confusing, if it works the  way you say it works. I mean,If the system tells me I have got this or that option available,then I expect it to be available.

Thanks.

[Karthick (Cloud Platform Support)]

Hello Michael, 

Thank you for providing more information. I noticed that you have already opened an issue tracker and I am currently investigating your project. And to avoid duplication of effort discussions in future will be held there.