Platform google cloud fastai

Returning to GCP

  1. Start instance
  2. if fail to start due to 'not enough resource', see troubleshoot
  3. log into console
  4. start instance
  5. SSH into instance:

    1. gcloud compute ssh --zone "us-west1-b" "my-fastai-instance" --project "project-name"
    2. gcloud compute ssh --zone=$ZONE jupyter@$INSTANCE_NAME -- -L 8080:localhost:8080
    3. gcloud compute ssh --zone=us-west1-b jupyter@my-fastai-instance -- -L 8080:localhost:8080
    4. go to browser: http://localhost:8080/tree
  6. Update course repo

  7. cd tutorials/fastai/course-v3
  8. git pull

  9. Update fastai library

  10. sudo /opt/anaconda3/bin/conda install -c fastai fastai

  11. scp securely copy files from gcp vm back to local: gcloud compute scp --recurse jupyter@my-fastai-instance-v20200528:~/tutorials/fastai/course-v3/nbs/dl1/lesson3-imdb-max-v20200528.ipynb /Users/max/Documents/ds-local/ds-fastai --zone "us-east1-b"

  12. When done, SHUT DOWN instance!!

Troubleshoot not enough resource

  1. Try to retry
  2. Create a new instance ```shell export IMAGE_FAMILY="pytorch-latest-gpu" # or "pytorch-latest-cpu" for non-GPU instances export ZONE="us-east1-b" #or try other zones us-west1-a/b, us-central1-c/f, us-east1-b/c export INSTANCE_NAME="my-fastai-instance-v20200528" # add a date for easy reference export INSTANCE_TYPE="n1-highmem-8" # budget: "n1-highmem-4"

# budget: 'type=nvidia-tesla-T4,count=1' gcloud compute instances create $INSTANCE_NAME \ --zone=$ZONE \ --image-family=$IMAGE_FAMILY \ --image-project=deeplearning-platform-release \ --maintenance-policy=TERMINATE \ --accelerator="type=nvidia-tesla-p100,count=1" \ --machine-type=$INSTANCE_TYPE \ --boot-disk-size=200GB \ --metadata="install-nvidia-driver=True" #\ #--preemptible #disable preemptible when unstable ```


  1. Google official document:
  2. Troubleshoot question on stackoverflow:
  3. fastai setup guide: