hiro/ray
1
0
Fork 0
mirror of https://github.com/vale981/ray synced 2025-03-06 02:21:39 -05:00
Code Issues Projects Releases Packages Wiki Activity Actions

[Serve] [Doc] Delete docs for removed automatic conda activation feature (#17832)

Browse source
This commit is contained in:
architkulkarni 2021-08-16 08:59:49 -07:00 • committed by GitHub
parent f2a3085ce2
commit e1ffc0fd73
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 3 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
doc/source/serve/core-apis.rst
View file

@ -235,7 +235,7 @@ Dependency Management
=====================
Ray Serve supports serving deployments with different (possibly conflicting)
python dependencies. For example, you can simultaneously serve one deployment
Python dependencies. For example, you can simultaneously serve one deployment
that uses legacy Tensorflow 1 and another that uses Tensorflow 2.
Currently this is supported on Mac OS and Linux using `conda <https://docs.conda.io/en/latest/>`_
@ -244,7 +244,8 @@ As with all other actor options, pass these in via ``ray_actor_options`` in
your deployment.
You must have a conda environment set up for each set of
dependencies you want to isolate. If using a multi-node cluster, the
desired conda environment must be present on all nodes.
desired conda environment must be present on all nodes. Also, the Python patch version
(e.g. 3.8.10) must be identical on all nodes (this is a requirement for any Ray cluster).
See :ref:`runtime-environments` for details.
Here's an example script. For it to work, first create a conda
@ -254,13 +255,6 @@ Python versions must be the same in both environments.
.. literalinclude:: ../../../python/ray/serve/examples/doc/conda_env.py
.. note::
If a conda environment is not specified, your deployment will be started in the
same conda environment as the client (the process creating the deployment) by
default. (When using :ref:`ray-client`, your deployment will be started in the
conda environment that the Serve controller is running in, which by default is the
conda environment the remote Ray cluster was started in.)
The dependencies required in the deployment may be different than
the dependencies installed in the driver program (the one running Serve API
calls). In this case, you should use a delayed import within the class to avoid