Fix Issue: “Asymmetric transaction rollback error” if commit callbacks throw exceptions

[waynetheisinger]

If a callback in commit throws an excecption, the calling plugin needs to be able to distinguish this from a unsuccessful commit, so that it doesn't attempt a rollback on an already commited transaction. (Which would cause an “Asymmetric transaction rollback" error)

Description

Fixed Issues (if relevant)

1. “Asymmetric transaction rollback error” if commit callbacks throw exceptions #6497: “Asymmetric transaction rollback error” if commit callbacks throw exceptions

Manual testing scenarios

Let a commit callback throw an exception

Example:

1. create a new search engine

    <type name="Magento\Search\Model\AdapterFactory">
        <arguments>
            <argument name="adapters" xsi:type="array">
                <item name="dummy" xsi:type="string">Magento\Framework\Search\Adapter\Mysql\Adapter</item>
            </argument>
        </arguments>
    </type>
   

2. Do not register a corresponding indexer handler for Magento\CatalogSearch\Model\Indexer\IndexerHandlerFactory

3. The commit callback from the fulltext indexer plugin will cause an exception "There is no such indexer handler: dummy"

4. Trigger this commit (for example, save a product)

Expected result

Exception from the commit callback is thrown and can be catched later (for ex. by PHPUnit)

Actual result

rollback() is called after commit() already has been called for the same transaction, resulting in an “Asymmetric transaction rollback error” exception when Magento tries to commit or rollback the last transaction on the stack.

Contribution checklist

• Pull request has a meaningful description of its purpose
• All commits are accompanied by meaningful commit messages
• All new or changed code is covered with unit/integration tests (if applicable)
• All automated tests passed successfully (all builds on Travis CI are green)

[maghamed]

Make sense for me.

Thanks for contribution

[maghamed]

I left a comment regarding this PR in the GitHub issue #6497

#6497 (comment)

[waynetheisinger]

@maghamed I have replied to your comment regarding this PR in the GitHub issue #6497

#6497 (my reply)

[waynetheisinger]

@maghamed - I've reverted the original changes and added the logging - so this is ready for another look.