Attempts to stabilize and improve speed of static checks #14332
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Member
Author
|
cc: @xinbinhuang |
potiuk
force-pushed
the
improve-static-checks-speed
branch
from
4d95126 to
5e8c9e4
Compare
Member
Author
|
The first results show that in self-hosted runner environment we can decrease pylint checks run from ~20 minutes to ~6 minutes when this one gets implemented. So I'd say - let's merge it quickly. I think stability will be improved as well as we again separate out test pylints from the "main" pylint and I think some of the recent stability issues were connected with tests accidentally being checked with "main" configuration. |
potiuk
force-pushed
the
improve-static-checks-speed
branch
from
5e8c9e4 to
f39b5f5
Compare
mik-laj
approved these changes
Feb 20, 2021
|
The PR most likely needs to run full matrix of tests because it modifies parts of the core of Airflow. However, committers might decide to merge it quickly and take the risk. If they don't merge it quickly - please rebase it to the latest master at your convenience, or amend the last commit of the PR, and push it with --force-with-lease. |
github-actions
bot
added
the
full tests needed
|
The Workflow run is cancelling this PR. Building images for the PR has failed. Follow the the workflow link to check the reason. |
potiuk
force-pushed
the
improve-static-checks-speed
branch
from
f39b5f5 to
d985734
Compare
turbaszek
approved these changes
Feb 20, 2021
Member
turbaszek
left a comment
turbaszek
left a comment
There was a problem hiding this comment.
pylintrc and pylintrc-tests
👏
|
The Workflow run is cancelling this PR. Building images for the PR has failed. Follow the the workflow link to check the reason. |
potiuk
force-pushed
the
improve-static-checks-speed
branch
from
d985734 to
91b2ca3
Compare
|
The Workflow run is cancelling this PR. Building images for the PR has failed. Follow the the workflow link to check the reason. |
potiuk
force-pushed
the
improve-static-checks-speed
branch
from
91b2ca3 to
a62b7ef
Compare
|
The Workflow run is cancelling this PR. Building images for the PR has failed. Follow the the workflow link to check the reason. |
potiuk
force-pushed
the
improve-static-checks-speed
branch
from
a62b7ef to
6265130
Compare
|
The Workflow run is cancelling this PR. Building images for the PR has failed. Follow the the workflow link to check the reason. |
potiuk
force-pushed
the
improve-static-checks-speed
branch
from
6265130 to
38e4f06
Compare
This change attempts to stabilize pylint checks. Since we have recently added self-hosted runners with multiple CPUS, seems that re-enabling parallel mode for pylint makes perfect sense as we will finally be able to use the parallelism and speed up static checks significantly. Previously the tests were run in single-processor mode in attempt to avoid random mistakes where different files were processed in different processes. This led to random pylint or mypy problems and aither false-positives or false negatives before especially when it came to circular dependencies. but since we are now past heavy refactoring, this should be no problem for future changes and occasional false positive/negative is less disruptive than long checks. The attempt is made to apply sorting order in the files passed to pylint. This should provide more stability in the results of running the tests in PR and in master. We had some custom pylint plugins that prevented using of pylint parallelism. For now we are giving up on one of the plugins (no asserts use) and we rely on committer's review on that (we have a rule in place to only use asserts in tests). The other plugin was replaced by coming back to separation of "main code" and "test code" and applying different rules to those - we have now two different configuration files|: pylintrc and pylintrc-tests to control settings for those two different cases. Mypy and flake8 have been parallelized at the level of pre-commits. By implementing those changes we are likely to speed up the tests on self-hosted runners 6x-8x times. Also caching of pre-commit environments (including the pre-commit installation) is improved. Previously we only cached the environment created by the pre-commit, but with this change we also cache the pre-commit installation in .local directory - this should save 1-2 minutes of the job.
potiuk
force-pushed
the
improve-static-checks-speed
branch
from
38e4f06 to
8fb7b3b
Compare
potiuk
added a commit
to potiuk/airflow
that referenced
this pull request
Feb 25, 2021
The apache#14332 change introduced --local flag for static checks and caching of the pre-commit virtualenv but the necessary PATH change was not added to `basic_static_checks.sh`. This PR fixes it.
potiuk
added a commit
that referenced
this pull request
Feb 25, 2021
The #14332 change introduced --local flag for static checks and caching of the pre-commit virtualenv but the necessary PATH change was not added to `basic_static_checks.sh`. This PR fixes it.
potiuk
added a commit
that referenced
this pull request
Mar 3, 2021
This change attempts to stabilize pylint checks. Since we have recently added self-hosted runners with multiple CPUS, seems that re-enabling parallel mode for pylint makes perfect sense as we will finally be able to use the parallelism and speed up static checks significantly. Previously the tests were run in single-processor mode in attempt to avoid random mistakes where different files were processed in different processes. This led to random pylint or mypy problems and aither false-positives or false negatives before especially when it came to circular dependencies. but since we are now past heavy refactoring, this should be no problem for future changes and occasional false positive/negative is less disruptive than long checks. The attempt is made to apply sorting order in the files passed to pylint. This should provide more stability in the results of running the tests in PR and in master. We had some custom pylint plugins that prevented using of pylint parallelism. For now we are giving up on one of the plugins (no asserts use) and we rely on committer's review on that (we have a rule in place to only use asserts in tests). The other plugin was replaced by coming back to separation of "main code" and "test code" and applying different rules to those - we have now two different configuration files|: pylintrc and pylintrc-tests to control settings for those two different cases. Mypy and flake8 have been parallelized at the level of pre-commits. By implementing those changes we are likely to speed up the tests on self-hosted runners 6x-8x times. Also caching of pre-commit environments (including the pre-commit installation) is improved. Previously we only cached the environment created by the pre-commit, but with this change we also cache the pre-commit installation in .local directory - this should save 1-2 minutes of the job. (cherry picked from commit 82cb041)
potiuk
added a commit
that referenced
this pull request
Mar 3, 2021
potiuk
added a commit
that referenced
this pull request
Mar 3, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This change attempts to stabilize pylint checks. Since we have
recently added self-hosted runners with multiple CPUS, seems that
re-enabling parallel mode for pylint makes perfect sense as we will
finally be able to use the parallelism and speed up static checks
significantly.
Previously the tests were run in single-processor mode in attempt
to avoid random mistakes where different files were processed in
different processes. This led to random pylint or mypy problems
and aither false-positives or false negatives before especially
when it came to circular dependencies. but since we are now past
heavy refactoring, this should be no problem for future changes
and occasional false positive/negative is less disruptive than
long checks.
The attempt is made to apply sorting order in the files passed
to pylint. This should provide more stability in the results
of running the tests in PR and in master.
We had some custom pylint plugins that prevented using of pylint
parallelism. For now we are giving up on one of the plugins
(no asserts use) and we rely on committer's review on that (we
have a rule in place to only use asserts in tests). The other
plugin was replaced by coming back to separation of "main code"
and "test code" and applying different rules to those - we have
now two different configuration files|: pylintrc and
pylintrc-tests to control settings for those two different cases.
Mypy and flake8 have been parallelized at the level of pre-commits.
By implementing those changes we are likely to speed up the
tests on self-hosted runners 6x-8x times.
Also caching of pre-commit environments (including the
pre-commit installation) is improved. Previously we only cached the
environment created by the pre-commit, but with this change we
also cache the pre-commit installation in .local directory - this
should save 1-2 minutes of the job.
^ Add meaningful description above
Read the Pull Request Guidelines for more information.
In case of fundamental code change, Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in UPDATING.md.