What this PR does / why we need it: This PR removes added "retrying" logic for gardenlinux.s3.S3Artifacts as the underlying boto3 implementation has an advanced and upstream supported feature already build-in.

I know about retry mechanism in boto. The idea here is to have a unifying interface for retries across the library, as we need this behavior not only for s3.

What this PR does / why we need it: This PR removes added "retrying" logic for gardenlinux.s3.S3Artifacts as the underlying boto3 implementation has an advanced and upstream supported feature already build-in.

I know about retry mechanism in boto. The idea here is to have a unifying interface for retries across the library, as we need this behavior not only for s3.

It would be better to feed unified timeout values than overwriting and actually having a high chance of multiplying the retry times applied to S3 connections.

What this PR does / why we need it: This PR removes added "retrying" logic for gardenlinux.s3.S3Artifacts as the underlying boto3 implementation has an advanced and upstream supported feature already build-in.

I know about retry mechanism in boto. The idea here is to have a unifying interface for retries across the library, as we need this behavior not only for s3.

It would be better to feed unified timeout values that overwriting and actually having a high chance of multiplying the retry times applied to S3 connections.

Got you. How about we set boto retries to 0 to avoid the multiplying effect?

Got you. How about we set boto retries to 0 to avoid the multiplying effect?

I would still prefer to use the boto3 way of handling retries as I see a high risk that we "break" upstream intended logic otherwise if we just add an overarching retry context manager to individual methods in our library and ignoring the effort that has been taken upstream.

Got you. How about we set boto retries to 0 to avoid the multiplying effect?

I would still prefer to use the boto3 way of handling retries as I see a high risk that we "break" upstream intended logic otherwise if we just add an overarching retry context manager to individual methods in our library and ignoring the effort that has been taken upstream.

Alright. I'm fine with that, but please don't remove the library. I see it can be useful in other places in this library.

Alright. I'm fine with that, but please don't remove the library. I see it can be useful in other places in this library.

I agree but we are not using it at the moment. As we use Git we can always return to the state where this library has been added once the functionality is needed.

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 87.62%. Comparing base (6ba28d5) to head (1efe214).
⚠️ Report is 2 commits behind head on main.

@@            Coverage Diff             @@
##             main     #209      +/-   ##
==========================================
- Coverage   87.66%   87.62%   -0.05%     
==========================================
  Files          33       33              
  Lines        1549     1543       -6     
==========================================
- Hits         1358     1352       -6     
  Misses        191      191              

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