Bundling kserving backend and client in the same pypi by jinchihe · Pull Request #250 · kserve/kserve

jinchihe · 2019-07-16T07:40:42Z

What this PR does / why we need it:
Develop python pip package for KFServing SDK, user can directly install kfserving sdk using pip install kfserving, and then call kfserving sdk apis to create/check/patch/delete kfservice.

Which issue(s) this PR fixes :
Fixes #246

Special notes for your reviewer:

Please confirm that if this PR changes any image versions, then that's the sole change this PR makes.

Release note:

Support installing KFServing Python SDK by `pip install kfserving`

This change is

jinchihe · 2019-07-18T03:10:43Z

The kfserving python SDK has been published to here: https://pypi.org/project/kfserving/

And user can install kfserving sdk from pip easily, and kfserving SDK works fine after installation.

(python36) [root@jinchi1 sdk]# pip install kfserving
Collecting kfserving
  Downloading https://files.pythonhosted.org/packages/71/7d/62da6cfcf79522351ac1d1327fbdcfdc7de84a75afc7596df210b69c8dab/kfserving-0.0.1.tar.gz
.....
Building wheels for collected packages: kfserving
  Building wheel for kfserving (setup.py) ... done
  Stored in directory: /root/.cache/pip/wheels/f1/8d/2f/c4b8db954491f4d997178c302acc3d4e66585899d25bb2e9db
Successfully built kfserving
Installing collected packages: kfserving
Successfully installed kfserving-0.0.1

jinchihe · 2019-07-18T03:11:58Z

/assign @yuzisun
Thank you in advance!

jinchihe · 2019-07-18T05:16:22Z

/retest

jinchihe · 2019-07-18T05:48:14Z

/retest

jinchihe · 2019-07-18T10:24:19Z

/retest

yuzisun · 2019-07-20T01:55:27Z

mostly looks good to me, do you currently publish to PyPi manually ?

jinchihe · 2019-07-23T08:10:00Z

mostly looks good to me, do you currently publish to PyPi manually ?

Yes, I do that manually, since there are only one command python setup.py sdist upload -r pypi to do that, and before running the CLI, we also need to configure the .pypirc for user-owned username/password, so I did not set it in script.

jinchihe · 2019-07-26T08:11:44Z

/hold

We are discussing more details in the #268 , may update once somethings decided. Thanks.

jinchihe · 2019-07-26T09:51:21Z

Hello Reviews:

From discussion and analysis from the ticket #268 , I update the PR changes, since the update is large, I would like to explain more for easy revewing. Overall, I merged the sdk into the python/kfserving, the especial changes are as following:

Merged python/kfserving/kfserving/__init__.py and sdk/kfserving/__init__.py
Merged the python/kfserving/setup.py and sdk/setup.py
Merged the README.md under python/kfserving/ and sdk
Moved test_server.py and test_storage.py under python/kfserving/kfserving/ to python/kfserving/test
Updated hack/Python-sdk/README.md to describe how to release kfserving SDK to Pypi.
Udpated kf_serving_client.py and test_kfservice_client.py to handle pylint problem (ignored some file for pylint since that generated, better to do not updates).
Ignored autogenerated below files but I think useless: .swagger-codegen-ignore .swagger-codegen/ .travis.yml git_push.sh test-requirements.txt tox.ini under sdk

There is a problem that the sdk-gen.sh will overwrite some file such current README.md in the future, that need to be updated that manually before check-in.

/cc @yuzisun @ellis-bigelow @animeshsingh

jinchihe · 2019-08-07T09:41:39Z

/hold cancel

animeshsingh · 2019-08-08T00:05:39Z

CC @yuzisun

ellistarn

Nice work!!

jinchihe · 2019-08-11T12:27:27Z

Rebased to resolve the conficts, but presubmit tests faild, that's caused by no space in automation env.

fatal: could not create leading directories of '/mnt/test-data-volume/kubeflow-kfserving-presubmit-e2e-250-7dd4267-5664-a27c/src/kubeflow
/kfserving': No space left on device

animeshsingh · 2019-08-12T02:15:24Z

There is a problem that the sdk-gen.sh will overwrite some file such current README.md in the future, that need to be updated that manually before check-in.

Can we make a comment on top of sdk-gen.sh to remind that folks need to replace README.md post the sdk generation?

Nice job @jinchihe

jinchihe · 2019-08-12T08:56:24Z

Can we make a comment on top of sdk-gen.sh to remind that folks need to replace README.md post the sdk generation?

@animeshsingh Great comments! Added notes in the scripts and README file for the script.

jinchihe · 2019-08-12T13:25:31Z

/retest

ellistarn · 2019-08-12T22:48:25Z

Nice!
/lgtm
/approve

k8s-ci-robot · 2019-08-12T22:48:33Z

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: ellis-bigelow

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

~~OWNERS~~ [ellis-bigelow]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

jinchihe · 2019-08-12T23:50:31Z

/retest

jinchihe · 2019-08-13T00:06:36Z

/retest

animeshsingh · 2019-08-13T00:13:31Z

/lgtm

animeshsingh · 2019-08-13T00:13:45Z

@yuzisun @ellis-bigelow final comments on this?

jinchihe · 2019-08-13T00:28:37Z

@yuzisun @ellis-bigelow final comments on this?

@animeshsingh The PR has been approved by Ellis, and That's OK from Dan, but the presummit testing already failed since timeout or some strange reasons...

/retest

* kfservice python sdk package * merge the sdk to kfserving

Motivation The Triton runtime can be used with model-mesh to serve PyTorch torchscript models, but it does not support arbitrary PyTorch models i.e. eager mode. KServe "classic" has integration with TorchServe but it would be good to have integration with model-mesh too so that these kinds of models can be used in distributed multi-model serving contexts. Modifications The bulk of the required changes are to the adapter image, covered by PR kserve/modelmesh-runtime-adapter#34. This PR contains the minimal controller changes needed to enable the support: - TorchServe ServingRuntime spec - Add "torchserve" to the list of supported built-in runtime types - Add "ID extraction" entry for TorchServe's gRPC Predictions RPC so that model-mesh will automatically extract the model name from corresponding request messages Note the supported model format is advertised as "pytorch-mar" to distinguish from the existing "pytorch" format that refers to raw TorchScript .pt files as supported by Triton. Result TorchServe can be used seamlessly with ModelMesh Serving to serve PyTorch models, including eager mode. Resolves kserve#63 Signed-off-by: Nick Hill <nickhill@us.ibm.com>

#### Motivation Support for TorchServe was added in kserve#250 and kserve/modelmesh-runtime-adapter#34. A test should be added for it as well. #### Modifications - Adds basic FVT for load/inference with a TorchServe MAR model using the native TorchServe gRPC API - Disables OVMS runtime and tests to allow TorchServe to be tested due to resource constraints #### Result Closes kserve#280 Signed-off-by: Rafael Vasquez <raf.vasquez@ibm.com>

…ent-updates/kserve-agent-28 Update kserve-agent-28 to fc7ef75

k8s-ci-robot added the do-not-merge/work-in-progress label Jul 16, 2019

k8s-ci-robot requested review from bhupc and gaocegege July 16, 2019 07:40

k8s-ci-robot added the size/M label Jul 16, 2019

johnugeorge reviewed Jul 16, 2019

View reviewed changes

Comment thread sdk/requirements.txt Outdated

jinchihe changed the title ~~[WIP] kfservice python sdk package~~ kfservice python sdk package Jul 18, 2019

k8s-ci-robot removed the do-not-merge/work-in-progress label Jul 18, 2019

k8s-ci-robot assigned yuzisun Jul 18, 2019

yuzisun reviewed Jul 20, 2019

View reviewed changes

Comment thread sdk/setup.py Outdated

yuzisun mentioned this pull request Jul 23, 2019

SDK Issues #263

Closed

k8s-ci-robot added do-not-merge/hold size/L and removed size/M labels Jul 26, 2019

k8s-ci-robot requested review from animeshsingh, ellistarn and yuzisun July 26, 2019 09:51

jinchihe changed the title ~~kfservice python sdk package~~ [WIP] Bundling kserving backend and client in the same pypi Jul 26, 2019

k8s-ci-robot added the do-not-merge/work-in-progress label Jul 26, 2019

animeshsingh reviewed Jul 30, 2019

View reviewed changes

Comment thread python/kfserving/setup.py Outdated

Comment thread python/kfserving/setup.py

Comment thread python/kfserving/setup.py

jinchihe mentioned this pull request Jul 31, 2019

add basic sdk api and sample notebook #245

Merged

k8s-ci-robot removed the size/L label Aug 6, 2019

jinchihe changed the title ~~[WIP] Bundling kserving backend and client in the same pypi~~ Bundling kserving backend and client in the same pypi Aug 7, 2019

k8s-ci-robot removed the do-not-merge/work-in-progress label Aug 7, 2019

k8s-ci-robot removed the do-not-merge/hold label Aug 7, 2019

ellistarn reviewed Aug 8, 2019

View reviewed changes

Comment thread hack/Python-sdk/README.md

Comment thread python/kfserving/README.md Outdated

kfservice python sdk package

3b147c1

merge the sdk to kfserving

bfcddf9

k8s-ci-robot assigned ellistarn Aug 12, 2019

k8s-ci-robot added the lgtm label Aug 12, 2019

k8s-ci-robot added the approved label Aug 12, 2019

k8s-ci-robot assigned animeshsingh Aug 13, 2019

k8s-ci-robot merged commit 2c52846 into kserve:master Aug 13, 2019

jinchihe deleted the update_setup_for_kservice_python_pkg branch August 13, 2019 00:54

jinchihe mentioned this pull request Aug 13, 2019

Discussion for bundling backend and client together for kfserving pypi #268

Closed

ellistarn pushed a commit to ellistarn/kfserving that referenced this pull request Jul 28, 2020

Bundling kserving backend and client in the same pypi (kserve#250)

ec949ae

* kfservice python sdk package * merge the sdk to kfserving

terrytangyuan pushed a commit to terrytangyuan/kserve that referenced this pull request May 13, 2024

Merge pull request kserve#250 from red-hat-data-services/rhtap/compon…

4c36dde

…ent-updates/kserve-agent-28 Update kserve-agent-28 to fc7ef75

Conversation

jinchihe commented Jul 16, 2019 • edited by jlewi Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

jinchihe commented Jul 18, 2019

Uh oh!

jinchihe commented Jul 18, 2019

Uh oh!

jinchihe commented Jul 18, 2019

Uh oh!

jinchihe commented Jul 18, 2019

Uh oh!

jinchihe commented Jul 18, 2019

Uh oh!

yuzisun commented Jul 20, 2019

Uh oh!

Uh oh!

jinchihe commented Jul 23, 2019

Uh oh!

jinchihe commented Jul 26, 2019

Uh oh!

jinchihe commented Jul 26, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

jinchihe commented Aug 7, 2019

Uh oh!

animeshsingh commented Aug 8, 2019

Uh oh!

ellistarn left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

jinchihe commented Aug 11, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

animeshsingh commented Aug 12, 2019

Uh oh!

jinchihe commented Aug 12, 2019

Uh oh!

jinchihe commented Aug 12, 2019

Uh oh!

ellistarn commented Aug 12, 2019

Uh oh!

k8s-ci-robot commented Aug 12, 2019

Uh oh!

jinchihe commented Aug 12, 2019

Uh oh!

jinchihe commented Aug 13, 2019

Uh oh!

animeshsingh commented Aug 13, 2019

Uh oh!

animeshsingh commented Aug 13, 2019

Uh oh!

jinchihe commented Aug 13, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

jinchihe commented Jul 16, 2019 •

edited by jlewi

Loading

jinchihe commented Jul 26, 2019 •

edited

Loading

jinchihe commented Aug 11, 2019 •

edited

Loading