[BEAM-12411] Update Tensorflow to version 2.5.0 & grpcio to 1.34.0

[aaltay]

Please add a meaningful description for your change here

---

Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily:

• Choose reviewer(s) and mention them in a comment (R: @username).
• Format the pull request title like [BEAM-XXX] Fixes bug in ApproximateQuantiles, where you replace BEAM-XXX with the appropriate JIRA issue, if applicable. This will automatically link the pull request to the issue.
• Update CHANGES.md with noteworthy changes.
• If this contribution is large, please file an Apache Individual Contributor License Agreement.

See the Contributor Guide for more tips on how to make review process smoother.

ValidatesRunner compliance status (on master branch)

Lang	ULR	Dataflow	Flink	Samza	Spark	Twister2
Go	---			---		---
Java
Python	---			---		---
XLang				---		---

Examples testing status on various runners

Lang	ULR	Dataflow	Flink	Samza	Spark	Twister2
Go	---	---	---	---	---	---	---
Java	---		---	---	---	---	---
Python	---	---	---	---	---	---	---
XLang	---	---	---	---	---	---	---

Post-Commit SDK/Transform Integration Tests Status (on master branch)

Go	Java	Python

Pre-Commit Tests Status (on master branch)

---	Java	Python	Go	Website	Whitespace	Typescript
Non-portable
Portable	---			---	---	---

See .test-infra/jenkins/README for trigger phrase, status and link of all Jenkins jobs.

GitHub Actions Tests Status (on master branch)

See CI.md for more information about GitHub Actions CI.

[codecov]

Codecov Report

Merging #14888 (81808d1) into master (79c1932) will decrease coverage by 0.03%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #14888      +/-   ##
==========================================
- Coverage   83.78%   83.75%   -0.04%     
==========================================
  Files         434      436       +2     
  Lines       58260    58491     +231     
==========================================
+ Hits        48816    48991     +175     
- Misses       9444     9500      +56     

Impacted Files	Coverage Δ
...eam/runners/interactive/caching/streaming_cache.py
...s/python/apache_beam/typehints/sharded_key_type.py
.../srcs/sdks/python/apache_beam/coders/coder_impl.py
...m/testing/benchmarks/nexmark/models/auction_bid.py
...n/apache_beam/runners/dataflow/dataflow_metrics.py
...y38/build/srcs/sdks/python/apache_beam/io/utils.py
...he_beam/testing/benchmarks/nexmark/nexmark_perf.py
.../python/apache_beam/io/gcp/resource_identifiers.py
...s/sdks/python/apache_beam/examples/avro_bitcoin.py
...e_beam/io/gcp/internal/clients/storage/__init__.py
... and 860 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 79c1932...81808d1. Read the comment docs.

[aaltay]

Run Portable_Python PreCommit

[aaltay]

Run PythonDocker PreCommit

[iemejia]

@aaltay It seems this requires to update also the grpcio version as a requisite too.

[aaltay]

@aaltay It seems this requires to update also the grpcio version as a requisite too.

You are right. Updated that too.

[aaltay]

Failing with:

13:55:01 RuntimeError: Could not retrieve licences for packages ['keras-nightly'] in Python3.8 environment. 
13:55:01  These licenses were not able to be pulled automatically.

Not sure how this change cause us to depend on keras-nightly. (@yifanmai - Do you have any idea?)

@iemejia - Changing to 2.4.2 might be easier but that is not a released version yet.

[yifanmai]

That is strange because I'd expect only tf-nightly to depend on keras-nightly.

[aaltay]

Run Python PreCommit

[aaltay]

That is strange because I'd expect only tf-nightly to depend on keras-nightly.

It turns out tensorflow==2.5.0 depends on keras-nightly ~=2.5.0.dev. I can reproduce it in a clean virtual environment. I will file a bug.

It was added here: tensorflow/tensorflow@d171d94

[iemejia]

That is strange because I'd expect only tf-nightly to depend on keras-nightly.

It turns out tensorflow==2.5.0 depends on keras-nightly ~=2.5.0.dev. I can reproduce it in a clean virtual environment. I will file a bug.

It was added here: tensorflow/tensorflow@d171d94

Can you please link the tensorflow issue for future reference. In the Java side we do not allow to include any SNAPSHOT dependency because of the lack of reproducibility. Is this ok to do on the python side?

[aaltay]

Run PythonDocker PreCommit

[aaltay]

That is strange because I'd expect only tf-nightly to depend on keras-nightly.

It turns out tensorflow==2.5.0 depends on keras-nightly ~=2.5.0.dev. I can reproduce it in a clean virtual environment. I will file a bug.
It was added here: tensorflow/tensorflow@d171d94

Can you please link the tensorflow issue for future reference. In the Java side we do not allow to include any SNAPSHOT dependency because of the lack of reproducibility. Is this ok to do on the python side?

I did not file a bug on their public issue tracker. The summary of the response was this was not intentional and it will be fixed in the next tensorflow release. If it helps, I can ask for a public bug for that. I do not think this is ok on python, and it was a bug.

On the license side, I think we are ok. Because even the tool was failing to parse, there is a license and apparently we have a file on our repo to manually track licenses for packages if the tooling we use fails.

[aaltay]

Run PythonDocker PreCommit

[aaltay]

@iemejia - tests are passing now. Let me know if this looks good.

[iemejia]

LGTM, It is ok if the tensorflow team is already aware of it.

[iemejia]

Thanks @aaltay . Can you please take care of cherry-picking this one for 2.31.0 since it is a security related upgrade.