kubeflow
/
notebooks
mirror of https://github.com/kubeflow/notebooks.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Add RWO_PVC_SCHEDULING env var to the Tensorboard Controller deployment (kubeflow/kubeflow#5266) 

* Add RWO_PVC_SCHEDULING env var to Tensorboard controller deployment

The value of the 'RWO_PVC_SCHEDULING' env var is set to "false" by
default. The user will be able to change the value of the env var
manually by modifying the 'config/manager/manager.yaml' file.

* Update README.md
This commit is contained in:
Konstantinos Andriopoulos 2020-08-31 18:12:21 +03:00 committed by GitHub
parent 8be6caebc5
commit a97b442e5b
2 changed files with 12 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
components/tensorboard-controller/README.md
View File

@ -39,7 +39,7 @@ Steps:
4. Run the controller locally: `make run`
If you want to enable the scheduling functionality for Tensorboard Controllers that use ReadWriteOnce PVCs as log storages, then set the `RWO_PVC_SCHEDULING` to `true` and run: `RWO_PVC_SCHEDULING="true" make run`
If you want to enable the scheduling functionality for Tensorboard servers that use ReadWriteOnce PVCs as log storages, then set the `RWO_PVC_SCHEDULING` to `true` and run: `RWO_PVC_SCHEDULING="true" make run`
## BUILD TENSORBOARD CONTROLLER IMAGE AND DEPLOY TO CLUSTER
@ -51,4 +51,11 @@ If you want to enable the scheduling functionality for Tensorboard Controllers t
4. Build and push the docker image: `make docker-build docker-push IMG=YOUR_IMAGE_NAME`
5. Deploy the Tensorboard controller: `make deploy IMG=YOUR_IMAGE_NAME`
5. Deploy the Tensorboard controller: `make deploy IMG=YOUR_IMAGE_NAME`
If you want to enable the scheduling functionality for Tensorboard servers that use ReadWriteOnce PVCs as log storages, then:
1. Change directories to `components/tensorboard-controller/config/manager`
2. Modify the `manager.yaml` file by navigating to the `deployment.spec.template.spec` field and manually setting the value of the `RWO_PVC_SCHEDULING` env var to `"true"` in the manager container.
3. Run: `make deploy IMG=YOUR_IMAGE_NAME`

3
components/tensorboard-controller/config/manager/manager.yaml
View File

@ -29,6 +29,9 @@ spec:
- --enable-leader-election
image: controller:latest
name: manager
env:
- name: RWO_PVC_SCHEDULING
value: "false"
resources:
limits:
cpu: 100m