Prevent AttributeError failure when the API attempts a database update retry after initially failing with an etag mismatch

[Jonny Rylands (jonnyry)]

If a Cosmos update fails due to an etag mismatch, and the retry succeeds - the calling code then fails due to no return value passed back.

The occured while two services were installing and both updating the firewall in the pipeline. Unfortunately I don't have the error log anymore as the test system is torn down.

[Copilot]

Pull request overview

This PR fixes a bug in the TRE API’s CosmosDB update retry path where an initial ETag mismatch followed by a successful retry could still cause the caller to fail due to a missing returned Resource. It also records the fix in the changelog and bumps the API patch version.

Changes:

• Propagate the successful retry result by returning the recursive try_update_with_retries(...) call result.
• Bump api_app patch version to 0.26.1.
• Add a BUG FIXES entry describing the retry/ETag issue.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.

File	Description
CHANGELOG.md	Adds a bug-fix entry for the ETag mismatch retry issue.
api_app/service_bus/helpers.py	Fixes the retry path to return the updated Resource from a successful retry attempt.
api_app/_version.py	Increments the API patch version to reflect the bug fix.

[github-actions]

Unit Test Results

724 tests   724 ✅  8s ⏱️
  1 suites    0 💤
  1 files      0 ❌

Results for commit 1a0fe9d.

♻️ This comment has been updated with latest results.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.

Suppressed comments (1)

api_app/service_bus/helpers.py:134

• The retry path is now correctly returning the result, but there’s no regression test covering the specific scenario this PR fixes: first Cosmos patch raises CosmosAccessConditionFailedError, then a subsequent retry succeeds and the successful Resource is returned to the caller (not None). Adding a test for the “fail once, succeed once” path would prevent this from regressing (e.g., extend api_app/tests_ma/test_service_bus/test_resource_request_sender.py to make patch_resource raise once then return (resource, template), and assert try_update_with_retries returns the resource and calls patch_resource twice).

    except CosmosAccessConditionFailedError as e:
        logger.warning(f"Etag mismatch for {resource_to_update_id}. Retrying.")
        if attempt_count < num_retries:
            return await try_update_with_retries(
                num_retries=num_retries,

[Jonny Rylands (jonnyry)]

Closing... work moved to PR #5025