Commit 55a08b61 authored by salbiach's avatar salbiach
Browse files

yaml and readme updated

parent 5f1e2792
......@@ -15,25 +15,37 @@ Or if you already have one you want to use:
Prepare your image. Dockerfile is in docker folder. Execute Makefile in case you need to build and push to docker repository everytime. Be aware that "PREFIX" in Makefile sohuld be changed accordingly to the repository you are using to obtain the image.
If you want to deploy k8s overriding existing one, first delete with command:
If you want to deploy k8s overriding existing one, first delete with command (in kubernetes folder):
`kubectl delete -f compss_deephealth.yaml`
Modify the number of worker replicas in line 69 to adapt it to your needs.
Go to kubernetes folder and execute:
`kubectl create -f compss_deephealth.yaml`
Wait for the pods to be initialized.
In case of need anything from the pod, you can get log:
`kubectl logs **POD NAME** -f`
You can also execute commands into pod's containers like this:
Enter the pod's container like:
`kubectl exec -it **POD NAME** -- /bin/bash` (in case pod only has one container)
Run:
`bash configure_compss.sh`
Modify runcompss.sh to your needs
Run:
`bash runcompss.sh`
In case of need anything from the pod, you can get log:
`kubectl logs **POD NAME** -f`
Both MNIST synchronous and asynchronous implementations are working correctly, calling by default the asynchronous one.
Other implementations can be found on the examples folder but further testing is required.
......@@ -28,7 +28,7 @@ spec:
type: DirectoryOrCreate
containers:
- name: dh-compss-master
image: registry.gitlab.bsc.es/ppc-bsc/software/deep-health-compss/compss-eddl:latest-compss-2.8-eddlgpu-1.1-python-3.6
image: registry.gitlab.bsc.es/ppc-bsc/software/deep-health-compss/compss-eddl:latest-compss-2.8-eddlgpu-1.1-python-3.6-v2
resources:
limits:
nvidia.com/gpu: 1 # requesting 1 GPU
......@@ -66,7 +66,7 @@ spec:
selector:
matchLabels:
app: compss
replicas: 16
replicas: 4
template:
metadata:
labels:
......@@ -84,7 +84,7 @@ spec:
type: DirectoryOrCreate
containers:
- name: dh-compss-worker
image: registry.gitlab.bsc.es/ppc-bsc/software/deep-health-compss/compss-eddl:latest-compss-2.8-eddlgpu-1.1-python-3.6
image: registry.gitlab.bsc.es/ppc-bsc/software/deep-health-compss/compss-eddl:latest-compss-2.8-eddlgpu-1.1-python-3.6-v2
resources:
limits:
nvidia.com/gpu: 1 # requesting 1 GPU
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment