Maintenance: latest deadline compatability issue #504
Comments
|
It can be possible from time to time that our dependencies are out of sync. To help with issue we can utilize python virtual environment. Without virtual environments, Python packages get installed into a global package site on the host. Each time you pip install ... a Python distributable (e.g. our DCC adaptors) into the global namespace, there is a chance of encountering a version conflict between the distributable's transitive dependencies and the dependencies of previously installed distributable. By installing each DCC into its own python environment these dependency conflicts can be avoided. Here is an example on how to achieve this setup: Now to access the binaries of the environment you will need to add the bin directory to PATH. To do this use your preferred method for setting PATH such as Now the Adaptor is accessible from the venv installation: |
|
Hello!
Thanks for the response and for taking a look at this.
I suppose this is was more of a query as to when the
`deadline-cloud-worker-agent` would have a release that supports the
breaking change to `deadline-cloud` going to.49
The deadline-cloud-worker-agent project requires deadline-cloud 0.48, while
all other projects in this domain have released support for 0.49.
Rather than make use of the latest greatest everywhere in my virtual env, I
had to roll back all other projects to the least common denominator...
which is this project...
Could it be that this project doesnt *actually* require deadline-cloud
0.49... it just hasn't been looked at for awhile so as to update its
requirements numbers?
That last version was released in oct for 0.48, while all other project
packages released something new in Nov for 0.49.
At any rate, looking forward to a release here, its there a way I can setup
notifications for that?
…On Wed, Dec 11, 2024 at 11:06 AM Charles Moore ***@***.***> wrote:
It can be possible from time to time that our dependencies are out of
sync. To help with issue we can utilize python virtual environment
<https://packaging.python.org/en/latest/guides/installing-using-pip-and-virtual-environments/#create-and-use-virtual-environments>.
Without virtual environments, Python packages get installed into a global
package site on the host. Each time you pip install ... a Python
distributable (e.g. our DCC adaptors) into the global namespace, there is a
chance of encountering a version conflict between the distributable's
transitive dependencies and the dependencies of previously installed
distributable. By installing each DCC into its own python environment these
dependency conflicts can be avoided.
Here is an example on how to achieve this setup:
python -m venv /opt/deadline/deadline-cloud-for-maya
source /opt/deadline/deadline-cloud-for-maya/bin/activate
pip install deadline-cloud-for-maya
exit
Now to access the binaries of the environment you will need to add the bin
directory to PATH.
To do this use your preferred method for setting PATH such as .bashrc or
.profile
export PATH=/opt/deadline/deadline-cloud-for-maya/bin:$PATH
Now the Adaptor is accessible from the venv installation:
MayaAdaptor --help
—
Reply to this email directly, view it on GitHub
<#504 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AU6BUETK4QALVTDUCI6XPEL2FCEMDAVCNFSM6AAAAABTKMEFVGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKMZWHA4DOMZTGY>
.
You are receiving this because you authored the thread.Message ID:
***@***.***
com>
|
This is not the case. The worker agent does depend on
I think the issue here is using a combined virtual environment for the worker agent and the other DCC adaptors. If each adaptor has a distinct virtual environment, the dependency closures for each application become isolated from one another and you no longer have to worry about conflicts. Have you considered creating a dedicated virtual environment per-DCC-adaptor as suggested by @moorec-aws above? |
|
Hi Josh, thanks for the reply.
I considered it, but it would require a bit of pipeline retooling on my
end, and I don't feel that is worth the effort at this time.
I can settle in with 0.48 compliant versions for now, until this update
comes available, as I cant imagine it would lag that far behind the others.
Cheers.
…On Thu, Dec 12, 2024 at 8:41 AM Josh Usiskin ***@***.***> wrote:
Could it be that this project doesnt *actually* require deadline-cloud
0.49... it just hasn't been looked at for awhile so as to update its
requirements numbers?
This is not the case. The worker agent does depend on deadline for
syncing input/output job attachments.
Rather than make use of the latest greatest everywhere in my virtual env, I
had to roll back all other projects to the least common denominator...
which is this project...
I think the issue here is using a combined virtual environment for the
worker agent and the other DCC adaptors. If each adaptor has a distinct
virtual environment, the dependency closures for each application become
isolated from one another and you no longer have to worry about conflicts.
Have you considered creating a dedicated virtual environment
per-DCC-adaptor as suggested by @moorec-aws
<https://github.com/moorec-aws> above?
—
Reply to this email directly, view it on GitHub
<#504 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AU6BUERICZSUJCU5W274JFD2FG4EVAVCNFSM6AAAAABTKMEFVGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKMZZGQ3DINJXGI>
.
You are receiving this because you authored the thread.Message ID:
***@***.***
com>
|
|
No problem, @embassytech and thanks for reaching out. The worker agent's pinned dependency version of To avoid this same issue coming up again in the future I'd recommend de-coupling virtual environments between Python applications. Best regards! |
|
Hi @embassytech, The worker agent version I am closing this issue now. Please feel free to re-open if you have any questions or need further clarity. Thanks! |
|
Fantastic. Thank you.
…On Mon, Dec 16, 2024 at 7:47 AM Josh Usiskin ***@***.***> wrote:
Hi @embassytech <https://github.com/embassytech>,
The worker agent version 0.27.5
<https://github.com/aws-deadline/deadline-cloud-worker-agent/releases/tag/0.27.5>
was released on Dec 14 which supports deadline versions 0.49.*.
I am closing this issue now. Please feel free to re-open if you have any
questions or need further clarity.
Thanks!
—
Reply to this email directly, view it on GitHub
<#504 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AU6BUETGJUOARGZ2AVPAOH32F3YZZAVCNFSM6AAAAABTKMEFVGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNBVHE4DCMZVHE>
.
You are receiving this because you were mentioned.Message ID:
***@***.***
com>
|
Description
Hi, I just did a
pip install --upgradefor the latestdeadline-cloudboto3and the various dcc submitter projects.Unfortunately, I was surprised to discover the latest version of this package is not compatible with all those.
$ pip install --upgrade deadline ... Using cached deadline-0.49.3-py3-none-any.whl (249 kB) Installing collected packages: deadline Attempting uninstall: deadline Found existing installation: deadline 0.48.9 Uninstalling deadline-0.48.9: Successfully uninstalled deadline-0.48.9 ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. deadline-cloud-worker-agent 0.27.4 requires deadline==0.48.*, but you have deadline 0.49.3 which is incompatible. Successfully installed deadline-0.49.3Will there be a new release of this that is shortly?
Sorry for the duplicate issue.
thanks.
Solution
release a version that supports deadline-clound 0.49.*
The text was updated successfully, but these errors were encountered: