Connect with docker session in examples #222

mkundu1 · 2022-03-24T06:02:00Z

No description provided.

mkundu1 · 2022-03-24T06:17:38Z

.github/workflows/ci.yml


+concurrency:
+  group: ${{ github.ref }}
+  cancel-in-progress: true


This will automatically cancel any in-progress run when pushed (https://docs.github.com/en/actions/using-workflows/workflow-syntax-for-github-actions#example-using-concurrency-to-cancel-any-in-progress-job-or-run)

dnwillia-work · 2022-03-29T02:35:21Z

examples/00-fluent/mixing_elbow_settings_api.py

+from ansys.fluent.core import examples
+
+if "PYFLUENT_RUN_EXAMPLES_WITH_DOCKER" in os.environ:
+    examples.download_file("mixing_elbow.pmdb", "pyfluent/mixing_elbow")


I guess in principle you could always download the files right?

These examples should be ready for public consumption and not contain any cryptic environment variables.

We need to find a way so that the doc build script launches a docker container before calling each example script. Within the example script we will only connect to the running server.

Or we can modify the launch_fluent command to first try to launch the Fluent docker container and connect to it. If the docker command/image is not found or if forced by an env var it will look for the local Fluent installation.

@mkundu1 Regarding your first sentence you can just add the docker calls to the nightly build CI:

https://github.202132.xyz.mcas.ms/pyansys/pyfluent/blob/main/.github/workflows/nightly-doc-build.yml

and the regular PR based CI.

Yes, you would have to modify launch_fluent to connect to it, makes sense. I've not stared at launch_apdl() but it must already be doing this right.

Regarding the file dependencies I think the point of the examples repo is to not have the files in the Fluent repo at all, so we should be removing them.

I think for luanch_fluent it's mostly there. Won't this work once the docker is spun up launch_fluent(start_instance=False). The container should be running locally and you need to just spin it up on a default port #, which should be open.

dnwillia-work · 2022-03-29T02:48:14Z

examples/00-fluent/mixing_elbow_settings_api.py

+    examples.download_file("mixing_elbow.scdoc", "pyfluent/mixing_elbow")
+
+    license_server = os.getenv("LICENSE_SERVER", "leblnxlic64.ansys.com")
+    subprocess.run(["docker", "run", "-d", "--rm", "-p", "63084:63084",


@mkundu1 Note that in the apdl CI pipeline they spin up docker as part of the CI.yml for the nightly doc builds:

https://github.com/pyansys/pymapdl/blob/51f4cb5f5f718cf0747929b59120aef48816464d/.github/workflows/nightly-doc-build.yml#L47

For the CI on a pull request they use this caching thing instead:

https://github.com/pyansys/pymapdl/blob/51f4cb5f5f718cf0747929b59120aef48816464d/.github/workflows/ci.yml#L147

I think what this is doing then is updating the examples once a day and every other time when the CI executes it uses the cached copy.

Pymapdl also runs the script .ci/start_mapdl.sh which spins up the docker as a part of unit testing and doc build. Their docs seems to be deployed nightly or when a tag is pushed ('Deploy' step).

OK, I missed that file shell script. I did not note that the CI scripts are using it anywhere. Let me see.

Yes, dev.fluentdocs.pyansys.com is updated nightly just like theirs, that's what our nightly CI yml is doing.

The released doc at fluentdocs.pyansys.com is updated when we tag, the update for this is in our ci.yml. We've not tagged yet so that's never ran.

The script seems to be used for the tests but looks to me like it contains identical commands to what is in their nightly CI script for the doc build. I'm not sure why they do this in two places. Probably to debug issues building the doc locally would be one guess (i.e. you can't run a ci.yml locally...)

License server should not be specified in these examples either.

What is the reason for needing to spin up new containers for each example? I don't understand. That sounds a bit unusable.

@dnwillia-work In standalone Fluent regression, we generally don't reuse the Fluent session between tests. The Fluent docker container starts with the Fluent session. So there we should use separate container for each tests (or the example scripts during the doc build). There can be issues like if the first script changes the container to solution mode, the next script cannot easily start from meshing mode.

I think this is normal usage scenario for containers - we can launch multiple of them treating each as isolated process.

That's a non-solution. You should be able to reset the container state. If you run 100s of unit tests and 100s of examples you'll spend the entire day restarting the service.

Currently we don't have an exposed way to reset the Fluent session. We need to look into having an API to do the same.

Yeah, I agree with Alex, so we would need that. Let's finish this PR for now, so that we have an example up and running, and then do that.

mkundu1 · 2022-03-29T08:42:14Z

The doc build is now successful and the Examples section is created in the doc artifact.

mkundu1 · 2022-03-29T15:51:54Z

.ci/start_fluent.py

+    license_server = os.environ["ANSYSLMD_LICENSE_FILE"]
+    port = os.environ["PYFLUENT_FLUENT_PORT"]
+
+    subprocess.run(["docker", "run", "--name", "fluent_server", "-d", "--rm",


These scripts are specific for CI runs. There is an assumption that the host OS is linux (-v assumes same unix path format for the container and host OS).

mkundu1 · 2022-03-29T15:54:22Z

doc/source/conf.py

+)
+
+
+def _start_or_stop_fluent_container(gallery_conf, fname, when):


runs before and after each example script

dnwillia-work · 2022-03-29T17:53:33Z

examples/00-fluent/mixing_elbow_settings_api.py

+import_filename = examples.download_file(
+    "mixing_elbow.pmdb", "pyfluent/mixing_elbow"
+)
+examples.download_file("mixing_elbow.scdoc", "pyfluent/mixing_elbow")


Why are you downloading this but only using the pmdb?

dnwillia-work

This is fine, but we have to come back to the reconnect issue.

…heat transfer in a mixing elbow

mkundu1 force-pushed the testing/mixing-elbow-with-docker branch 3 times, most recently from d6fdd1e to 2292b55 Compare March 24, 2022 06:14

mkundu1 commented Mar 24, 2022

View reviewed changes

mkundu1 force-pushed the testing/mixing-elbow-with-docker branch 6 times, most recently from 5a282ba to 7e9e52c Compare March 27, 2022 10:26

mkundu1 mentioned this pull request Mar 27, 2022

Run pytest with Fluent docker image #200

Closed

mkundu1 force-pushed the testing/mixing-elbow-with-docker branch 7 times, most recently from 69a0f79 to 4178c53 Compare March 28, 2022 07:57

mkundu1 requested a review from ypatel-qa March 28, 2022 15:49

dnwillia-work reviewed Mar 29, 2022

View reviewed changes

mkundu1 force-pushed the testing/mixing-elbow-with-docker branch 5 times, most recently from 04474ee to 1d95874 Compare March 29, 2022 05:30

mkundu1 requested review from seanpearsonuk and sujal-tipnis March 29, 2022 08:40

seanpearsonuk approved these changes Mar 29, 2022

View reviewed changes

mkundu1 commented Mar 29, 2022

View reviewed changes

mkundu1 force-pushed the testing/mixing-elbow-with-docker branch from 01e3ec8 to 7d77bfe Compare March 29, 2022 15:53

mkundu1 commented Mar 29, 2022

View reviewed changes

mkundu1 force-pushed the testing/mixing-elbow-with-docker branch from 7d77bfe to 261877e Compare March 29, 2022 17:16

dnwillia-work reviewed Mar 29, 2022

View reviewed changes

dnwillia-work approved these changes Mar 29, 2022

View reviewed changes

mkundu1 force-pushed the testing/mixing-elbow-with-docker branch from 261877e to 1c6e2ac Compare March 30, 2022 00:15

mkundu1 mentioned this pull request Mar 30, 2022

Add API to reset Fluent session to reuse docker container across multiple tests. #238

Closed

mkundu1 force-pushed the testing/mixing-elbow-with-docker branch from 1c6e2ac to 910c600 Compare March 30, 2022 00:54

ypatel-qa and others added 2 commits March 30, 2022 06:27

Adding an example demonstrating Fluent using Python: Fluid flow and …

67b92b8

…heat transfer in a mixing elbow

Connect with docker session in examples

4ed9899

mkundu1 force-pushed the testing/mixing-elbow-with-docker branch from 910c600 to 4ed9899 Compare March 30, 2022 00:58

mkundu1 merged commit 5022614 into main Mar 30, 2022

mkundu1 deleted the testing/mixing-elbow-with-docker branch March 30, 2022 01:19

mkundu1 mentioned this pull request Mar 30, 2022

Adding an example demonstrating Fluent using Python: Fluid flow and … #199

Closed

		)


		def _start_or_stop_fluent_container(gallery_conf, fname, when):

Connect with docker session in examples #222

Connect with docker session in examples #222

Uh oh!

Conversation

mkundu1 commented Mar 24, 2022

Uh oh!

mkundu1 Mar 24, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mkundu1 Mar 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dnwillia-work Mar 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dnwillia-work Mar 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dnwillia-work Mar 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dnwillia-work Mar 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dnwillia-work Mar 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mkundu1 commented Mar 29, 2022

Uh oh!

mkundu1 Mar 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dnwillia-work left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

mkundu1 Mar 24, 2022 •

edited

Loading

mkundu1 Mar 29, 2022 •

edited

Loading

dnwillia-work Mar 29, 2022 •

edited

Loading

dnwillia-work Mar 29, 2022 •

edited

Loading

dnwillia-work Mar 29, 2022 •

edited

Loading

dnwillia-work Mar 29, 2022 •

edited

Loading

dnwillia-work Mar 29, 2022 •

edited

Loading

mkundu1 Mar 29, 2022 •

edited

Loading