search

CONABIO / kube_sipecam

k8s processing documentation for SiPeCaM project
MIT License
0 stars 1 forks source link

It's not necessary to have two deployments of kale-jupyterlab-kubeflow_0.4.0_1.14.0 #8

Open palmoreck opened 4 years ago

palmoreck commented 4 years ago

It was seen after doing tests that is not necessary to distinguish between having next line:

https://github.com/CONABIO/kube_sipecam/blob/master/deployments/audio/kale-jupyterlab-kubeflow_0.4.0_1.14.0_tf.yaml#L35

and don't have it in deployment:

https://github.com/CONABIO/kube_sipecam/blob/master/deployments/audio/kale-jupyterlab-kubeflow_0.4.0_1.14.0_tf_cpu.yaml#L32

At least using the example for torch:

https://github.com/CONABIO/kube_sipecam_playground/tree/issue-1/audio/notebooks/dockerfiles/tf_kale/0.4.0_1.14.0_tf/cifar10

the kubeflow+kale run was successful

So I either could delete file

https://github.com/CONABIO/kube_sipecam/blob/master/deployments/audio/kale-jupyterlab-kubeflow_0.4.0_1.14.0_tf_cpu.yaml

or use this file to compile notebook via kale and avoid having problems in kubernetes for not finding nodes with gpu's (because stablishing inside limits block the paremeter nvidia.com/gpu: 1 causes this message)

palmoreck commented 3 years ago

Check:

https://github.com/NVIDIA/k8s-device-plugin#running-gpu-jobs

It says:


    WARNING: if you don't request GPUs when using the device plugin with NVIDIA images all the GPUs on the machine will be exposed inside your container.