ray
ray copied to clipboard
[RLLib] Critical Tensorflow CVE - CVE-2023-25664
What happened + What you expected to happen
Please see below critical CVE found in ray-ml 2.10.0
image:
SEVERITY IMPACTED PACKAGE FIXED VERSIONS CVE SCORE
Critical pypi://tensorflow:2.11.0 [2.11.1] CVE-2023-25664 9.8
Versions / Dependencies
The physical location:
(base) ray@kuberay-head-5z5vd:/$ pip show tensorflow
Name: tensorflow
Version: 2.11.0
Summary: TensorFlow is an open source machine learning framework for everyone.
Home-page: https://www.tensorflow.org/
Author: Google Inc.
Author-email: [email protected]
License: Apache 2.0
Location: /home/ray/anaconda3/lib/python3.8/site-packages
Requires: absl-py, astunparse, flatbuffers, gast, google-pasta, grpcio, h5py, keras, libclang, numpy, opt-einsum, packaging, protobuf, setuptools, six, tensorboard, tensorflow-estimator, tensorflow-io-gcs-filesystem, termcolor, typing-extensions, wrapt
Required-by: dopamine-rl, recsim
(base) ray@kuberay-head-5z5vd:/$
Reproduction script
NA
Issue Severity
High: It blocks me from completing my task.
min version should be 2.11 or higher now > finalize and review on Thu than close ticket cc @thomasdesr
doing a quick search brings this as part of the rllib dir path @simonsays1980 @sven1977 can one of you cut a PR to upgrade to TF latest (or at least 2.11.1) as @sercanCyberVision reported so we can close this CVE vuln?
@sven1977 @simonsays1980 please follow up. Thanks
@anyscalesam @zhe-thoughts Apologies for the delay - my Anyscale account got deleted, so I had to search actively on GitHub for triage issues.
Yes, this an issue mentioned already somewhere else. We take car of this.
thanks - when do you think you can submit a PR so we can merge into the next Ray weekly release @simonsays1980 ?
Sorry for the delay, the actual RLlib is NOT requiring this 2.11.0 version anymore. RLlib shares the exact same requirements as all other ML libraries through here.
What it could be is one of the rllib_contrib
algos, which we stopped maintaining (and froze dependencies for). Some of these algos are pinned to tf 2.11.0. ... I'll provide a PR to try upgrading all these to 2.11.1 ...