Support multi-platform storages for Tensorboard

[IronPan]

Currently tensorboard pod mounts the GCP service accounts and can access to GCS.

pipelines/frontend/server/k8s-helper.ts

Lines 46 to 79 in 24d69cb

	const pod = {
	kind: 'Pod',
	metadata: {
	generateName: 'tensorboard-',
	},
	spec: {
	containers: [{
	args: [
	'tensorboard',
	'--logdir',
	logdir,
	],
	image: 'tensorflow/tensorflow',
	name: 'tensorflow',
	ports: [{
	containerPort: 6006,
	}],
	env: [{
	name: 'GOOGLE_APPLICATION_CREDENTIALS',
	value: '/secret/gcp-credentials/user-gcp-sa.json'
	}],
	volumeMounts: [{
	mountPath: '/secret/gcp-credentials',
	name: 'gcp-credentials',
	}],
	}],
	volumes:[{
	name: 'gcp-credentials',
	secret:{
	secretName: 'user-gcp-sa',
	},
	}],
	},
	};

The credential should be a configurable parameter that user specifies, to read data from various storage, S3, PV, etc.

[Jeffwan]

@IronPan

Any idea how could we improve this manifest? The most straightforward way is check logdir protocol?

1. for gcs, s3 and minio, pod can mount cloud provider specific environment and secrets
2. for local disk, we need pvc name and mountPath?

[sarahmaddox]

@IronPan Can we close this issue, after the merge of the above PRs? We have a note in the docs that we can remove when this issue and #1700 are closed.

[IronPan]

/close

[k8s-ci-robot]

@IronPan: Closing this issue.

In response to this:

/close

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.

[IronPan]

@sarahmaddox thanks! close it for now