@arunoda I’ve tested this out https://github.com/arunoda/fastai-shell#switching-zones and I keep switching zones and I get the following errors for last 30minutes
ERROR: (gcloud.compute.instances.create) Could not fetch resource:
I’ve been using another GPU provider and I wanted to try GCP out…
Please help
This seems like a common issue since you don’t have GPU quota on that given region.
You can change it by visiting https://console.cloud.google.com/iam-admin/quotas and increase quota for GPUs you need.
Try searching metric for “gpu”.
If there’s something like “gpus-all-regions” increase that too.
how to stop jupyter noteoobk and re-start again?
Read here: https://github.com/arunoda/fastai-shell#getting-started
You can run fastai stop and fastai start again.
Is this starting and stopping instance? Is there a way to just stop start jupyter without restarting instance?
Could you explain your use case?
If you do this, you will need to pay even jupyter is not running.
Type this: sudo systemctl stop jupyter
Thank you. I wanted to restart jupyter as I had some weird out of memory issues for very small batch size like 4 and image size 256px using v100.
Ah cool. Makes sense.
You could try restarting Kernal from Jupyter as well.
Not sure you already did it or not.
Yeah. Restarted from within jupyter but the issue was happening again so wanted to do ‘hard’ reset. Thanks!
I’m using fastai-shell and it’s working really well for me.
Now I need to also be able to run Keras/ Tensorflow on my machine. Could anybody please indicate to me what is the best way to do it?
Hi arunoda,
Can it be that something has changed in the scripts that fastai-shell runs?
This morning, I tried bash update_fastai.sh, as I do every morning, but received a CorruptedEnvironmentError. No way to proceed.
After 60 min of googling, I came to the conclusion that I should start over, so fastai destroy (waited until boot disk and VM were gone), and then fastai create for a new boot disk and VM.
However, the script ends with:
`Get:178 http://ppa.launchpad.net/graphics-drivers/ppa/ubuntu bionic/main amd64 nvidia-kernel-common-396 amd64
- https://repo.anaconda.com/pkgs/pro/linux-64
- https://repo.anaconda.com/pkgs/pro/noarch
To search for alternate channels that may provide the conda package you'relooking for, navigate to
[https://anaconda.org](https://anaconda.org/)
and use the search bar at the top of the page.
Connection to 35.247.29.0 closed.`
This leaves my account is the state where fastai-boot-1 is listed under disk, but also under VM? SSH-ing into the boot disk finds that only anaconda3 is installed, no fastai material.
I’m unable to proceed further now. Any hints for this?
Output for fastai version:
0.2.0
Thank you!
Help please!
I’m still unable to use Keras/ Tensorflow in GCP.
Could any body give me some hint whether it’s possible to use this in the fastai-shell or not? Or do I need to create a totally separate environment to use Keras?
I really appreciate any help!
Updated to use the latest version of fastai and pytorch.
This should work as expected now on.