Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Ignore name of index when fetching settings #823

Merged
merged 1 commit into from
Sep 26, 2024

Conversation

tienne-B
Copy link
Contributor

Description

This commit fixes a situation where an index cannot be updated through this client as the index is aliased, with the client pointing to the alias.

As the GET /<index>/_settings request will only ever return the settings for the specified index (through the alias), it would only have one key, so the name of the key would not matter. We can pop the key to get the settings object for the index through the alias.

Describe what this change achieves.

Issues Resolved

Closes #822

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

Copy link

codecov bot commented Sep 18, 2024

Codecov Report

Attention: Patch coverage is 20.00000% with 8 lines in your changes missing coverage. Please review.

Project coverage is 70.34%. Comparing base (ba715b9) to head (e7532db).
Report is 63 commits behind head on main.

Files with missing lines Patch % Lines
opensearchpy/_async/helpers/index.py 20.00% 4 Missing ⚠️
opensearchpy/helpers/index.py 20.00% 4 Missing ⚠️
Additional details and impacted files
@@            Coverage Diff             @@
##             main     #823      +/-   ##
==========================================
- Coverage   71.95%   70.34%   -1.61%     
==========================================
  Files          91      113      +22     
  Lines        8001     8896     +895     
==========================================
+ Hits         5757     6258     +501     
- Misses       2244     2638     +394     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

Copy link
Member

@dblock dblock left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What's the value of _name vs. what it is after the change here?
Can you please add a test?

@tienne-B tienne-B force-pushed the patch-1 branch 2 times, most recently from 263b708 to 2ed7b6e Compare September 18, 2024 16:43
@tienne-B
Copy link
Contributor Author

Thank you @dblock for your quick response! The value of removing self._name is that saving an index previously wouldn't work if OS-py accesses it through an alias (and using an alias seems to be a good practice for downtime prevention). All the requests that are triggered by the .save() method can work transparently through an index, however, when making the GET /<alias>/_settings request, the dict response key is the name of the true index, rather than the alias. This causes the KeyError. However, the data we're extracting doesn't depend on the name, so it'd be safe to disregard and pop the key/value.

If it's wanted, I could add a check that the alias only points to one index to raise a more specific error. Would that be appreciated?

I've added a test, but I'm a little skeptical of the coverage report as it's based off a 6-month-old commit.

Thank you!

@tienne-B tienne-B requested a review from dblock September 18, 2024 16:56
@dblock
Copy link
Member

dblock commented Sep 19, 2024

If it's wanted, I could add a check that the alias only points to one index to raise a more specific error. Would that be appreciated?

Possibly, but doesn't have to block this PR. Give it a shot after I merge this!

Copy link
Member

@dblock dblock left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This all makes sense.

I think you have to do the same for the async code (see https://github.com/opensearch-project/opensearch-py/blob/main/opensearchpy/_async/helpers/index.py#L308). Add a test for that too.

@tienne-B
Copy link
Contributor Author

I've added support for async, the checks for multiple indices, and testing for these.

Copy link
Member

@dblock dblock left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good!

I have some small language asks for the error. Also check that the tests do run because CodeCov seems unhappy and complains that tests are missing which seems odd.

@@ -34,7 +34,7 @@
from opensearchpy import OpenSearch
from opensearchpy.exceptions import ConnectionError

OPENSEARCH_URL = os.environ.get("OPENSEARCH_URL", "https://localhost:9200")
OPENSEARCH_URL = os.environ.get("OPENSEARCH_URL", "http://localhost:9200")
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What's the reason for this change? AFAIK it should always be https.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh sorry, didn't mean to commit that! I had noticed that the tests did not run, and had to uncomment these list values to get the tests to run:

ignores = [
"test_opensearchpy/test_server/",
"test_opensearchpy/test_server_secured/",
"test_opensearchpy/test_async/test_server/",
"test_opensearchpy/test_async/test_server_secured/",
]

I am a little hesitant to commit such changes as I don't have much context as to why this has been done and if it'd break anything. Also, the base commit for Codecov is quite old.

Copy link
Member

@dblock dblock Sep 23, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ugh this looks like a mess. I opened #826 with a high level idea of what we should do here (maybe someone wants to help later, wink wink).

For this PR I'd like to ensure the tests you wrote actually run (and pass) in CI. Is this the case?

Copy link
Contributor Author

@tienne-B tienne-B Sep 25, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The tests do pass locally with the files un-ignored (although this assertion is somewhat empty)

On GitHub Actions they would be skipped as there is no OpenSearch service for the tests to use.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

That's the whole point, integration tests should run on GHA in one of the workflows.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is still here, do you want to revert this line?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reverted, sorry.

opensearchpy/_async/helpers/index.py Outdated Show resolved Hide resolved
@tienne-B tienne-B force-pushed the patch-1 branch 3 times, most recently from d0b4bed to 2edcaf3 Compare September 25, 2024 02:02
@tienne-B tienne-B requested a review from dblock September 25, 2024 02:09
@dblock
Copy link
Member

dblock commented Sep 25, 2024

@tienne-B so we can't merge this without a passing CI, help out with whatever the right answer is to get it to pass and ensuring that the tests you have added do run in CI, please?

I debugged the test failure but couldn't find a solution. You can rebase with the workaround in #828 for now (or maybe you know how to fix that failure?).

@tienne-B
Copy link
Contributor Author

@dblock Rebased.

Copy link
Member

@dblock dblock left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There's still that http, want to fix it?

@@ -34,7 +34,7 @@
from opensearchpy import OpenSearch
from opensearchpy.exceptions import ConnectionError

OPENSEARCH_URL = os.environ.get("OPENSEARCH_URL", "https://localhost:9200")
OPENSEARCH_URL = os.environ.get("OPENSEARCH_URL", "http://localhost:9200")
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is still here, do you want to revert this line?

This commit fixes a situation where an index cannot be updated through this client as the index is aliased, with the client pointing to the alias.

As the `GET /<index>/_settings` request will only ever return the settings for the specified index (through the alias), it would only have one key, so the name of the key would not matter. We can pop the key to get the settings object for the index through the alias.

Signed-off-by: Étienne Beaulé <[email protected]>
@dblock dblock merged commit 1b0440a into opensearch-project:main Sep 26, 2024
32 of 34 checks passed
@dblock
Copy link
Member

dblock commented Sep 26, 2024

good work merci @tienne-B !

@tienne-B tienne-B deleted the patch-1 branch September 26, 2024 18:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[BUG] Cannot update aliased index through Index.save()
2 participants