#71 Simplify accelerators config #90
Conversation
|
Thanks for your pull request. It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). 📝 Please visit https://cla.developers.google.com/ to sign. Once you've signed, please reply here (e.g.
|
|
Hi @wbuchwalter. Thanks for your PR. I'm waiting for a kubernetes member to verify that this patch is reasonable to test. If it is, they should reply with I understand the commands that are listed here. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. I understand the commands that are listed here. |
|
I signed it! |
|
CLAs look good, thanks! |
|
/ok-to-test |
README.md
Outdated
|
|
||
| 1. Deploy the operator | ||
|
|
||
| For non-RBAC enabled clusters: | ||
| ``` | ||
| CHART=https://storage.googleapis.com/tf-on-k8s-dogfood-releases/latest/tf-job-operator-chart-latest.tgz | ||
| helm install ${CHART} -n tf-job --wait --replace | ||
| helm install ${CHART} -n tf-job --wait --replace --set cloud=<gce or azure> |
README.md
Outdated
| ``` | ||
|
|
||
| For RBAC-enabled clusters: | ||
| ``` | ||
| CHART=https://storage.googleapis.com/tf-on-k8s-dogfood-releases/latest/tf-job-operator-chart-latest.tgz | ||
| helm install ${CHART} -n tf-job --wait --replace --set rbac.install=true | ||
| helm install ${CHART} -n tf-job --wait --replace --set rbac.install=true cloud=<gce or azure> |
README.md
Outdated
|
|
||
| The TfJob controller can be configured with a list of volumes that should be mounted from the host into the container | ||
| to make GPUs work. Here's an example [ControllerConfig](https://github.com/tensorflow/k8s/blob/master/pkg/spec/controller.go): | ||
| For **GCE** |
README.md
Outdated
| to make GPUs work. Here's an example [ControllerConfig](https://github.com/tensorflow/k8s/blob/master/pkg/spec/controller.go): | ||
| For **GCE** | ||
| ``` | ||
| helm install ${CHART} -n tf-job --wait --replace --set cloud=gce |
README.md
Outdated
| ``` | ||
|
|
||
| If the cluster is not hosted on GCE or Azure, you will need specify a custom configuration. | ||
| To do so edit `${CHART}\custom-config.yaml` with your desired settings. |
There was a problem hiding this comment.
I think editing the chart is undesirable because it requires the user download/unpack the chart.
What if instead of passing in a file to helm, the user has to create the config map manually. e.g.
kubectl create configmap tf-job-operator-config --from-file=path/to/bar
helm install ${CHART} -n tf-job --wait --replace --set cloud=none
if cloud=none then the helm chart doesn't include the tf-job-operator-config defined in the template.
There was a problem hiding this comment.
Done.
Slight difference, is that you don't need to specify any value for cloud. As long as it it neither gke nor azure the chart will not create a ConfigMap.
cloudvalue that acceptsgkeorazureand will create the correspondingConfigMapcustom-config.yamlfile at the chart's root allowing users to specify a custom config. Ifcloudis not passed,custom-config.yamlwill be used instead.Remarks:
config-fileflag, because in order to import a file in a Helm template, this file has to be at the root of the chart. Not doing so will result in a silent error (ConfigMap will just not be created).Since this provides a bad user experience, I chose to instead direct users to modify
custom-config.yamlthat will be already present at the the root of the cart.