CODEWITHSUNDEEP
google-dl-platform
user1425706
user1425706

Reputation: 195

Deep Learning VM fails to deploy via UI - image resource not found - typo in the image url

I've tried deploying the marketplace solution Deep Learning VM (Google Click to Deploy) using TF2.0 with a GPU. I'm doing this through the UI to select the zone and other instance options.

As soon as I deploy however and get taken to the Deployment Manager screen, I see the following error:

jupyterlab-eu-w-4c-vm: {"ResourceType":"compute.v1.instance","ResourceErrorCode":"400","ResourceErrorMessage":{"code":400,"errors":[{"domain":"global","message":"Invalid value for field 'resource.disks[0].initializeParams.sourceImage': 'https://www.googleapis.com/compute/v1/projects/click-to-deploy-images/global/images/tf-2-0-cu100-experimental-20190821'. The referenced image resource cannot be found.","reason":"invalid"}],"message":"Invalid value for field 'resource.disks[0].initializeParams.sourceImage': 'https://www.googleapis.com/compute/v1/projects/click-to-deploy-images/global/images/tf-2-0-cu100-experimental-20190821'. The referenced image resource cannot be found.","statusMessage":"Bad Request","requestPath":"https://compute.googleapis.com/compute/v1/projects/jupyterlab-instance/zones/europe-west4-c/instances","httpMethod":"POST"}}

The key being that the image resource cannot be found at that url:

https://www.googleapis.com/compute/v1/projects/click-to-deploy-images/global/images/tf-2-0-cu100-experimental-20190821

I searched for the available images on the cloud shell:

@cloudshell:~ (jupyterlab-instance)$ gcloud compute images list --project click-to-deploy-images --no-standard-images --uri | grep tf-2-0-cu100
https://www.googleapis.com/compute/v1/projects/click-to-deploy-images/global/images/tf-2-0-cu100--experimental-20190821

Notice, the URL is different, there is an extra "-" in the image name as compared to what the deployment script is trying to fetch:

tf-2-0-cu100-experimental-20190821
tf-2-0-cu100--experimental-20190821

This looks like an unintentional typo.

My question though, is how can I go about deploying this VM? Is there a way I can modify the deployment script that the UI generates before deploying or do I need to do the whole deployment via the CLI to add in the extra "-"?

Is there a way I can raise this to get someone to fix the typo? I presume this would be preventing anyone trying to deploy a TensorFlow 2 GPU instance via the UI tools using the Deep Learning VM.

Thanks for your help.

Upvotes: 2

Views: 759

Answers (2)

FlashDD
FlashDD

Reputation: 438

I encountered the same problem. that VM will not deploy with TF 2.0 version because the boot image URL is messed up it seems. it's not related to zone (I've tried deploying without GPUs and in different zones, it won't work)

One solution is to deploy the image with an instance directly (See documentation 1)

gcloud compute instances create $INSTANCE_NAME \
  --zone=$ZONE \
  --image-family=tf2-latest-gpu \(I used cpu the one but this one seems to fit)
  --image-project=deeplearning-platform-release \
  --accelerator=count=1,type=nvidia-tesla-k80

Add any options desired (GPU,etc...).

You can get help for the command with 

gcloud compute instances create --help

To list all available images, use

gcloud compute images list --project deeplearning-platform-release --no-standard-images

Upvotes: 1

an1lam
an1lam

Reputation: 649

I had a very similar problem and it turned out that I was trying to deploy a GPU model in a zone in which it wasn't supported. Take a look here to see whether the GPU type you're using is supported in "europe-west4-c". For example, if you're using a K80, then it's not available in that zone (see screenshot below).

k80 supported regions

Upvotes: 0

Related Questions