[go: nahoru, domu]

Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Support Workload Identity Federation authentication in GFile #57104

Open
lgeiger opened this issue Aug 11, 2022 · 4 comments
Open

Support Workload Identity Federation authentication in GFile #57104

lgeiger opened this issue Aug 11, 2022 · 4 comments
Assignees
Labels
comp:core issues related to core part of tensorflow stat:awaiting tensorflower Status - Awaiting response from tensorflower type:feature Feature requests

Comments

@lgeiger
Copy link
Contributor
lgeiger commented Aug 11, 2022
Click to expand!

Issue Type

Feature Request

Source

binary

Tensorflow Version

2.9.0

Custom Code

No

OS Platform and Distribution

Ubuntu 20:04

Mobile device

No response

Python version

3.9

Bazel version

No response

GCC/Compiler version

No response

CUDA/cuDNN version

No response

GPU model and memory

No response

Current Behaviour?

Workload Identity Federation is the new and preferred authentication mechanism to GCS on CI systems where long lived service account keys can pose a security risk according to https://github.com/google-github-actions/auth.

It would be great if TensorFlow gfile would be able to utilise this authentication mechanism as well.

Standalone code to reproduce the issue

A full reproduction can be found at https://github.com/google-github-actions/auth/issues/210

Relevant log output

No response

@google-ml-butler google-ml-butler bot added the type:feature Feature requests label Aug 11, 2022
@tilakrayal tilakrayal added the comp:core issues related to core part of tensorflow label Aug 12, 2022
@sachinprasadhs sachinprasadhs added the stat:awaiting tensorflower Status - Awaiting response from tensorflower label Aug 12, 2022
@carlthome
Copy link
Contributor

Any update on this from Googlers? This is a pretty tricky annoyance in our CI/CD workflows on GitHub Actions to Vertex and Dataflow, since gcloud just works while TensorFlow Datasets (tfds CLI and tfds.load) does not without jumping through hurdles.

@sachinprasadhs
Copy link
Contributor

Hi,

If this is issue is related to any security vulnerability of Tensorflow, could you please report it in the proper channel so that this will be taken on priority.
For more details on how to report these, please refer https://github.com/tensorflow/tensorflow/security/policy#vulnerabilities-in-tensorflow

@fathom-nathan
Copy link

cc @sachinprasadhs
Is there any update here either from Google or acceptable workaround found by the community? Workload Identity Federation has been in GA for 4+ years and is GCP's recommended pattern for impersonating service accounts vs. using service account keys. Pulling from GCP's own documentation for "Best practices for managing service account keys":

Service account keys can become a security risk if not managed carefully. You should choose a more secure alternative for authentication whenever possible. The main threats are related to service account keys are:

  • Credential leakage: Service account keys might inadvertently end up in places where they are not supposed to be stored. A bad actor can use a leaked service account key to authenticate and gain a foothold in your environment.
  • Privilege escalation: If a bad actor gets access to a poorly secured service account key, they might be able to use the key to escalate their privileges.
  • Information disclosure: Service account keys might inadvertently disclose confidential metadata.
  • Non-repudiation: By authenticating using a service account key and letting the service account carry out operations on their behalf, a bad actor might conceal their identity and actions.

@derekheld
Copy link
derekheld commented May 1, 2024

cc @sachinprasadhs Is there any update here either from Google or acceptable workaround found by the community? Workload Identity Federation has been in GA for 4+ years and is GCP's recommended pattern for impersonating service accounts vs. using service account keys. Pulling from GCP's own documentation for "Best practices for managing service account keys":

Even better Google has now gone even farther introducing direct auth, allowing you to fully bring your own identity to authenticate directly with GCP APIs instead of using an impersonated service account at all. Also recommended by Google over service account impersonation. That's certainly not supported here either.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
comp:core issues related to core part of tensorflow stat:awaiting tensorflower Status - Awaiting response from tensorflower type:feature Feature requests
Projects
None yet
Development

No branches or pull requests

6 participants