Return errors for insert_rows_json exceptions (#21080) #28091

ajdub508 · 2023-08-22T10:22:16Z

Would like to get feedback on this approach. Marking the MR as draft because more tests would need to be added and I haven't run existing tests yet, either.

The purpose of this change is to get exceptions caught and re-raised here to output rows to failed_rows rather than just causing a pipeline error. Outputting to failed_rows tag is necessary to avoid streaming pipelines that infinitely retry messages that result in these pipeline errors.

These exceptions cause infinite retries because this self.bigquery_wrapper.insert_rows call ends up raising an exception rather than returning errors due to the exception raised by self.gcp_bq_client.insert_rows_json that prevents the errors list from being instantiated. The exception will bubble up (raised here) to the _flush_batch method where it is unhandled and will produce a pipeline error.

Getting those rows to the failed_rows tag may require doing some things that may or may not be acceptable:

The exception has to be packaged up with the required fields
A reason has to be assigned to these errors. I went with 'invalid' to make sure that these types of errors are not retried
All rows in a batch are output to the failed_rows tag. Looking back at this comment on BEAM-12362, I think the original intent of raising these exceptions was to get the rows into the failed_rows tag and it seems likely that most/all rows in a batch would be affected by the types of exceptions being caught there.

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

Mention the appropriate issue in your description (for example: addresses #123), if applicable. This will automatically add a link to the pull request in the issue. If you would like the issue to automatically close on merging the pull request, comment fixes #<ISSUE NUMBER> instead.
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.

To check the build health, please visit https://github.com/apache/beam/blob/master/.test-infra/BUILD_STATUS.md

GitHub Actions Tests Status (on master branch)

See CI.md for more information about GitHub Actions CI or the workflows README to see a list of phrases to trigger workflows.

ajdub508 · 2023-08-25T09:30:31Z

R: @liferoad - would you be able to provide some feedback or suggest a reviewer?

github-actions · 2023-08-25T09:31:45Z

Stopping reviewer notifications for this pull request: review requested by someone other than the bot, ceding control

chamikaramj · 2023-08-28T21:22:28Z

@liferoad could you check or suggest a reviewer ?

liferoad · 2023-08-28T23:00:36Z

R: @Abacn

Sorry, this was lost in my notifications.
Is this PR ready to be reviewed since it is in a draft?

ajdub508 · 2023-08-28T23:20:22Z

@liferoad @Abacn - Just looking for early feedback on the approach. Provided some background in this comment.

liferoad · 2023-08-28T23:39:56Z

@liferoad @Abacn - Just looking for early feedback on the approach. Provided some background in this comment.

I see. Thanks.

R: @ahmedabu98

liferoad · 2023-08-28T23:45:10Z

sdks/python/apache_beam/io/gcp/bigquery_tools.py

-      raise
-    except HttpError as e:
+      # Package exception up with required fields
+      # Set reason to 'invalid' to consider these execptions as RetryStrategy._NON_TRANSIENT_ERRORS


If these errors are retry-able, the new changes will not be able to insert these rows.

Thanks @liferoad, I'll take a closer look at that. Let me know if I'm off on any of this, but my thinking had been - as the code currently stands the ClientError, GoogleAPICallError, and HttpError exceptions never get a chance to be retried anyway. So this doesn't take away a chance to be retried, just makes sure the rows can be captured in failed_rows and provide a way to disposition the message and ack a pubsub message.

The reason I had thought that the ClientError, GoogleAPICallError, and HttpError exceptions never get a chance to be retried is:

the exceptions gets re-raised in _insert_all_rows here

the exceptions is not caught by insert_rows here

and the _flush_batch method here isn't catching them either

End result of all that for ClientError, GoogleAPICallError, and HttpError exceptions is that the exception results in a pipeline error rather than producing a usable errors list, and RetryStrategy never gets a chance to be evaluated.

My thought was that it would be better to at least route those rows to the failed rows tag where users can choose what to do with them and avoid issues I've seen with pubsub messages that are never acked.

Shall we fix the code path to retry these errors (at least the errors are retry-able) instead of returning them as the error PColl?

@liferoad - I'm giving that some thought, will report back

Thanks. In general, for any batch job, Dataflow will retry the failed work items 4 times; for streaming jobs, it will retry forever. So if one error is retry-able, we could just raise the exception and let Dataflow handle it.

So if one error is retry-able, we could just raise the exception and let Dataflow handle it

IIUC this is the current behavior already. When this exception is raised, Dataflow sees it as a failed work item and retries it (batch 4 times, streaming infinitely).

ahmedabu98

Thank you for taking this on @ajdub508! I've seen users running into this problem frequently. Dataflow streaming retries failed work items infinitely so in the worst of cases they end up with a stuck pipeline.

I like this approach of throwing a batch in the DLQ if the insert call fails, it's better than hopelessly retrying and at least users can decide what to do with the failed rows in a downstream stage. However, I'd like to do it so that we still retry when it's appropriate to.

ahmedabu98 · 2023-09-01T13:00:53Z

sdks/python/apache_beam/io/gcp/bigquery_tools.py

+      # Set reason to 'invalid' to consider these execptions as RetryStrategy._NON_TRANSIENT_ERRORS
+      error = {
+        'message': e.message,
+        'reason': 'invalid'


I'd prefer a more granular approach than setting all these exceptions to invalid. Are we sure these errors will always be non-transient? Perhaps we can have an if-tree that sets an appropriate reason based on the error code. Or maybe better to pass in the actual error reason and update _NON_TRANSIENT_ERRORS to include more reasons.

For example the code 400 indicates an invalid error so we can say that's non-transient. But if we get something like a 500 or 503 indicates a temporary error and BQ suggests retrying. More info in this error messages documentation.

I'll look at adding to _NON_TRANSIENT_ERRORS, that's a good idea.

ahmedabu98 · 2023-09-01T13:17:10Z

sdks/python/apache_beam/io/gcp/bigquery_tools.py

-      raise
-    except HttpError as e:
+      # Package exception up with required fields
+      # Set reason to 'invalid' to consider these execptions as RetryStrategy._NON_TRANSIENT_ERRORS


So if one error is retry-able, we could just raise the exception and let Dataflow handle it

IIUC this is the current behavior already. When this exception is raised, Dataflow sees it as a failed work item and retries it (batch 4 times, streaming infinitely).

ahmedabu98 · 2023-09-01T13:22:03Z

sdks/python/apache_beam/io/gcp/bigquery_write_it_test.py

+  def test_big_query_write_insert_not_found_errors(self):
+    """
+    Test that NotFound errors returned by beam.io.WriteToBigQuery
+    contain both the failed rows amd the reason for it failing.


nit

Suggested change

contain both the failed rows amd the reason for it failing.

contain both the failed rows and the reason for it failing.

Thanks for the feedback, @ahmedabu98. I think we could raise 429, 500, and 503 errors to let the pipeline errors be retried indefinitely and route all the others to the failed rows tag. This would be consistent with errors defined as transient in API Core Retry and AIP-194

Those retryable errors would all be caught in the first except clause. The second clause catching HttpError type doesn't contain anything that is typically considered transient and they'll go straight to failed rows.

Thinking about this some more along with the suggestion to add to _NON_TRANSIENT_ERRORS, should be able to use the response reasons to let RetryStrategy do its thing, may not need to raise and cause a pipeline error.

ajdub508 · 2023-09-09T12:31:02Z

Restructured tests to cover the 2 types of issues that can arise from insert_rows_json calls:

errors list is returned from the function call with error reasons described here
an exception is thrown and caught with error reasons as described in api-core exceptions

Separate tests added for each retry strategy for both of those types of issues. Will cleanup any linter or test issues, incorporate any additional feedback, and get ready to move out of draft status next.

codecov · 2023-09-09T12:43:45Z

Codecov Report

Merging #28091 (2589f22) into master (08a9767) will increase coverage by 0.01%.
Report is 16 commits behind head on master.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master   #28091      +/-   ##
==========================================
+ Coverage   72.24%   72.26%   +0.01%     
==========================================
  Files         684      684              
  Lines      100952   100953       +1     
==========================================
+ Hits        72929    72949      +20     
+ Misses      26447    26428      -19     
  Partials     1576     1576

Flag	Coverage Δ
python	`82.86% <100.00%> (+0.02%)`	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed	Coverage Δ
sdks/python/apache_beam/io/gcp/bigquery_tools.py	`75.00% <100.00%> (+0.03%)`	⬆️

... and 11 files with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

ajdub508 · 2023-09-13T10:50:11Z

@liferoad @ahmedabu98 Moving this out of draft status, ready for review.

I did not change handling of HttpErrors because they are different from GoogleAPICallError exceptions. Exceptions from apitools.base.py.exceptions do not provide a reason value. I'm thinking we probably don't want to maintain a mapping from HttpError codes into usable reasons and the apitools library is deprecated, too.

ahmedabu98 · 2023-09-13T14:55:29Z

Run Python 3.8 PostCommit

ajdub508 · 2023-09-14T09:42:59Z

@ahmedabu98 The test failed but looks like the failures were unrelated.

I still need to add to CHANGES.md. Should I add under Bugfixes since the #21080 issue is tagged as a bug or is there a chance this should be considered a breaking change for anyone depending on error handling logic?

ahmedabu98

Thanks for the comments documenting the test cases! was really helpful when following along. I have a few clarifying questions there and a couple requests:

Can we keep the types of exceptions we currently have? We wouldn't want to lose all that coverage.
I see that the reasons added to _NON_TRANSIENT_ERRORS may be geared towards GRPC errors? Can we add a few reasons that BQ would return to us? for example notFound (this particular one is mentioned in WriteToBigQuery ignores insert_retry_strategy on HttpErrors #21080) or accessDenied

ahmedabu98 · 2023-09-13T14:03:49Z

sdks/python/apache_beam/io/gcp/bigquery_tools.py

+  # Values below may be found in reasons provided either in an
+  # error returned by a client method or by an http response as
+  # defined in google.api_core.exceptions
+  _NON_TRANSIENT_ERRORS = {
+      'invalid',
+      'invalidQuery',
+      'notImplemented',
+      'Moved Permanently',
+      'Not Modified',
+      'Temporary Redirect',
+      'Resume Incomplete',
+      'Bad Request',
+      'Unauthorized',
+      'Forbidden',
+      'Not Found',
+      'Method Not Allowed',
+      'Conflict',
+      'Length Required',
+      'Precondition Failed',
+      'Not Implemented',
+      'Bad Gateway',
+      'Gateway Timeout',
+  }


Thanks for adding this comprehensive list! I'm a little unfamiliar with a few of these so will need another pair of eyes to check.

From a cursory search though, it appears "bad gateway" may be a transient error? Same with "gateway timeout". I would defer to https://cloud.google.com/bigquery/docs/error-messages for a good list to take from.

Also error reasons from BQ seem to appear in camel case format (e.g. notFound instead of Not Found).
Is it different for GRPC errors?

Thanks for adding this comprehensive list! I'm a little unfamiliar with a few of these so will need another pair of eyes to check.

From a cursory search though, it appears "bad gateway" may be a transient error? Same with "gateway timeout". I would defer to https://cloud.google.com/bigquery/docs/error-messages for a good list to take from.

I see what you mean about some of the one I added possibly being transient, we certainly could take some of the 5XX errors out of that list to broaden it a bit.

I went narrow, initially, based on the narrow set of api core exceptions considered transient in api core retry here, where it's only 429 Too Many Requests, 500 Internal Server Error, and 503 Service Unavailable that are considered transient. Other than those 3, all remaining 4XX, 5XX, and 3XX errors are not considered transient.

Let me know if you think we should remove some 5XX exceptions from the list or if we should stick with just retrying 429, 500, and 503.

Also error reasons from BQ seem to appear in camel case format (e.g. notFound instead of Not Found). Is it different for GRPC errors?

Posted this in a different comment but will include here to make it easy to follow in this thread -

There are 2 types of error reasons in that list:

Reasons returned in the error list response of the self.gcp_bq_client.insert_rows_json call. Those are in camel case and are the ones found here in this doc - https://cloud.google.com/bigquery/docs/error-messages.

Reasons returned by the GoogleAPICalError exceptions. Those follow the HTTP standard reason format. A 404, for example will return a reason of Not Found, which is verified with this integration test.

Yeah let's be cautious which errors we label as "non-transient", because that could break many current pipelines that rely on Beam to successfully retry these errors. Previously these GRPC errors would be raised and the runner can retry the whole bundle.

I would lean towards only listing the errors we are sure to be non-transient.

Gotcha, I'll review that and trim that list down a bit, seems like retrying more of the 5XX errors would make sense.

Added this commit which removed a few exceptions based on our conversation and a review of the bq error table:

Errors that don't seem to be relevant:

3XX Errors - Moved Permanently, Not Modified, Temporary Redirect, Resume Incomplete

Some 4XX Errors - Method Not Allowed, Length Required, Precondition Failed

Errors that should be retried:

409 Conflict

502 Bad Gateway

504 Gateway Timeout

sdks/python/apache_beam/io/gcp/bigquery_test.py

ahmedabu98 · 2023-09-14T15:58:12Z

sdks/python/apache_beam/io/gcp/bigquery_test.py

-          exception_type=requests.exceptions.ConnectionError,
-          error_message='some connection error'),
+          exception_type=exceptions.TooManyRequests if exceptions else None,
+          error_reason='Too Many Requests', # not in _NON_TRANSIENT_ERRORS
+          streaming=False),
      param(
-          exception_type=requests.exceptions.Timeout,
-          error_message='some timeout error'),
+          exception_type=exceptions.Forbidden if exceptions else None,
+          error_reason='Forbidden', # in _NON_TRANSIENT_ERRORS
+          streaming=False),
      param(
-          exception_type=ConnectionError,
-          error_message='some py connection error'),
+          exception_type=exceptions.TooManyRequests if exceptions else None,
+          error_reason='Too Many Requests', # not in _NON_TRANSIENT_ERRORS
+          streaming=True),
      param(
-          exception_type=exceptions.BadGateway if exceptions else None,
-          error_message='some badgateway error'),


Likewise, these exceptions are important to keep. There was a large effort last year to add wide exception testing coverage to make sure these exceptions are handled properly.

I'd say the same with the rest of the current exception tests that we have. You can maybe try integrating those exceptions with your tests

Thanks, will add back.

Looking closer at this - I can add these back, but I still think they might be redundant. I think the tests in this change cover the possible code paths. This is my line of thinking, let me know where I'm off track:

All of the existing exception tests are mocking google.cloud.bigquery.Client.insert_rows_json

There 2 scenarios we need to handle when calling that method with regard to errors/exceptions:

the method returns an error list response here

the method raises a GoogleAPICallError exception here

In each of those scenarios, a different path can be taken based on whether the reason is in _NON_TRANSIENT_ERRORS, so the tests in this change exercise both transient/non-transient to cover both paths.

The tests are re-organized in this change to try to make it clear which path is being tested, with test_insert_rows_json_exception_* tests for the first scenario and test_insert_rows_json_errors_* tests for the 2nd scenario. Also exercising retry strategies and transient/non-transient errors for each scenario to cover those paths for each scenario.

I don't mean to be a pain on this and certainly may be missing something, so I will defer to project committers on this, just wanted to see if I can understand what I'm missing.

You're certainly right that fundamentally we should be focusing on those two code paths. I agree with not bringing those tests back, but maybe the exceptions themselves should be spread out across your tests. For example, we can replace one of the many exceptions.TooManyRequests in your test cases with exceptions.Timeout, and in another test case replace it with exceptions.BadGateway etc. So the tests you wrote can stay where they are, but just the parameters can change to include a wider variety of exceptions.

Although these exceptions ultimately take the same code path in the end, the variety of exceptions tested can guard against future unwanted changes to _NON_TRANSIENT_ERRORS. (e.g. someone making Timeout a non-transient error)

Thanks, that's a great point and good idea, I'll switch up some of the exceptions in those tests.

Added this commit which adds a variety of exceptions and stops setting the streaming option in the tests. I added test cases for the test_insert_rows_json_exception_retry_on_transient_error test to cover all expected retriable exceptions and all listed non-transient exceptions.

ahmedabu98 · 2023-09-14T20:21:06Z

sdks/python/apache_beam/io/gcp/bigquery_test.py

      param(
-          exception_type=requests.exceptions.Timeout,
-          error_args=('some timeout error', )),
+          insert_response=[
+              [{
+                  'index': 0, 'errors': [{
+                      'reason': 'invalid'
+                  }]
+              }],
+          ],
+          failed_rows=['value1'],
+          streaming=True),


I see this is the same test case as the first one in this sequence. What is the significance of streaming=True?

I noticed that there were some separate tests, some with streaming true/false since the behavior with pipeline errors is different. I added them to demonstrate/check behavior for streaming true/false. Let me know if you think that's unnecessary, I can remove one or the other.

Do you remember which tests?
I think it should be safe to remove them, the same code path is taken either way. I also don't see the connector checking streaming besides for validation reasons.

The old test_insert_all_unretriable_errors_streaming test used to set opt.streaming = True.

I will take those out, agree that they aren't needed.

Fixed with this commit

ajdub508 · 2023-09-15T09:33:43Z

Thanks for the comments documenting the test cases! was really helpful when following along. I have a few clarifying questions there and a couple requests:

Can we keep the types of exceptions we currently have? We wouldn't want to lose all that coverage.

I see that the reasons added to _NON_TRANSIENT_ERRORS may be geared towards GRPC errors? Can we add a few reasons that BQ would return to us? for example notFound (this particular one is mentioned in WriteToBigQuery ignores insert_retry_strategy on HttpErrors #21080) or accessDenied

I kept the original 3, 'invalid', 'invalidQuery', 'notImplemented', and have those at the top of the list.

There are 2 types of error reasons in that list:

Reasons returned in the error list response of the self.gcp_bq_client.insert_rows_json call. Those are in camel case and are the ones found here in this doc - https://cloud.google.com/bigquery/docs/error-messages.
Reasons returned by the GoogleAPICalError exceptions. Those follow the HTTP standard reason format. A 404, for example will return a reason of Not Found, which is verified with this integration test.

I can definitely add some errors from the bq docs to the list, wouldn't hurt anything, let me know if you think we should go that route.

I think a lot of them probably weren't included originally because they'll actually surface a GoogleAPICallError exception and wouldn't be returned in the error list response of the self.gcp_bq_client.insert_rows_json call. When that happens, like with the 404 example, we won't see the bq camel case notFound, we'll end up reading the Not Found reason from the GoogleAPICallError exception that we're catching instead. That's why this integration test works the way it does, ending up with a Not Found reason.

ahmedabu98 · 2023-09-15T11:10:15Z

I think a lot of them probably weren't included originally because they'll actually surface a GoogleAPICallError exception and wouldn't be returned in the error list response of the self.gcp_bq_client.insert_rows_json call

Ahh I see, that clarifies it for me. Thank you!

ahmedabu98

Some exceptions are still missing (e.g. ConnectionError, Timeout, DeadlineExceeded), but I won't block this PR on that.

LGTM, thank you @ajdub508!

ajdub508 · 2023-09-18T15:08:56Z

Thanks for the approval, @ahmedabu98, should I add a CHANGE.md entry and squash commits now?

ahmedabu98 · 2023-09-18T18:53:49Z

I think this could be added as a bug fix on CHANGES.md, and sure you can squash commits now. Let me know when you're ready and I can merge

johnjcasey · 2023-09-18T18:57:41Z

sdks/python/apache_beam/io/gcp/bigquery_tools.py

+      'Bad Request',
+      'Unauthorized',
+      'Forbidden',
+      'Not Found',


Not found could be transient, in that it could show up later. I don't think we should treat it as transient though

Right, I think we catch that earlier when validating the create disposition. In Java SDK we also tell the user to make sure the table is created before processing the writes. And if the table gets dropped halfway thru I think it makes sense to treat it as non-transient

ajdub508 · 2023-09-19T10:36:48Z

@ahmedabu98 Squashed commits, added CHANGES.md entry, and added a couple tests (here and here) for the exceptions you mentioned. Ready to be merged.

ahmedabu98 · 2023-09-19T16:16:23Z

Run Python 3.9 PostCommit

ahmedabu98 · 2023-09-19T19:05:30Z

Tests are passing, will merge now.

Thanks for sticking this one out @ajdub508!

github-actions bot added python io gcp labels Aug 22, 2023

ajdub508 changed the title ~~Return errors for insert_rows_json execptions (#21080)~~ Return errors for insert_rows_json exceptions (#21080) Aug 22, 2023

ajdub508 force-pushed the python-insert-exception-handling branch from 941cdb3 to 834d9a9 Compare August 22, 2023 10:23

ajdub508 mentioned this pull request Aug 22, 2023

WriteToBigQuery ignores insert_retry_strategy on HttpErrors #21080

Closed

liferoad reviewed Aug 28, 2023

View reviewed changes

ahmedabu98 reviewed Sep 1, 2023

View reviewed changes

ajdub508 force-pushed the python-insert-exception-handling branch 3 times, most recently from d1a7db7 to b5b863d Compare September 9, 2023 12:23

ajdub508 force-pushed the python-insert-exception-handling branch 2 times, most recently from 10aaa70 to dbb6573 Compare September 13, 2023 09:01

ajdub508 marked this pull request as ready for review September 13, 2023 10:48

ahmedabu98 reviewed Sep 14, 2023

View reviewed changes

ajdub508 force-pushed the python-insert-exception-handling branch from 7c11461 to 65d206d Compare September 16, 2023 11:50

ahmedabu98 approved these changes Sep 18, 2023

View reviewed changes

johnjcasey reviewed Sep 18, 2023

View reviewed changes

Return errors for insert_rows_json exceptions (apache#21080)

2589f22

ajdub508 force-pushed the python-insert-exception-handling branch from 65d206d to 2589f22 Compare September 19, 2023 10:14

ahmedabu98 merged commit 0f72792 into apache:master Sep 19, 2023
40 checks passed

	contain both the failed rows amd the reason for it failing.
	contain both the failed rows and the reason for it failing.

Return errors for insert_rows_json exceptions (#21080) #28091

Return errors for insert_rows_json exceptions (#21080) #28091

Conversation

ajdub508 commented Aug 22, 2023 • edited Loading

GitHub Actions Tests Status (on master branch)

ajdub508 commented Aug 25, 2023

github-actions bot commented Aug 25, 2023

chamikaramj commented Aug 28, 2023

liferoad commented Aug 28, 2023

ajdub508 commented Aug 28, 2023

liferoad commented Aug 28, 2023

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ahmedabu98 left a comment

Choose a reason for hiding this comment

ahmedabu98 Sep 1, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ajdub508 Sep 2, 2023 • edited Loading

Choose a reason for hiding this comment

ajdub508 commented Sep 9, 2023

codecov bot commented Sep 9, 2023 • edited Loading

Codecov Report

ajdub508 commented Sep 13, 2023

ahmedabu98 commented Sep 13, 2023

ajdub508 commented Sep 14, 2023

ahmedabu98 left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ahmedabu98 Sep 15, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ajdub508 commented Sep 15, 2023

ahmedabu98 commented Sep 15, 2023

ahmedabu98 left a comment

Choose a reason for hiding this comment

ajdub508 commented Sep 18, 2023

ahmedabu98 commented Sep 18, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ajdub508 commented Sep 19, 2023

ahmedabu98 commented Sep 19, 2023

ahmedabu98 commented Sep 19, 2023

ajdub508 commented Aug 22, 2023 •

edited

Loading

ahmedabu98 Sep 1, 2023 •

edited

Loading

ajdub508 Sep 2, 2023 •

edited

Loading

codecov bot commented Sep 9, 2023 •

edited

Loading

ahmedabu98 left a comment •

edited

Loading

ahmedabu98 Sep 15, 2023 •

edited

Loading

ahmedabu98 commented Sep 18, 2023 •

edited

Loading