DataNucleus JIRA is now in read-only mode. Raise any new issues in GitHub against the plugin that it applies to. DataNucleus JIRA will remain for the foreseeable future but will eventually be discontinued
Issue Details (XML | Word | Printable)

Key: NUCCORE-738
Type: Task Task
Status: Closed Closed
Resolution: Fixed
Priority: Major Major
Assignee: Unassigned
Reporter: Andy Jefferson
Votes: 0
Watchers: 0
Operations

If you were logged in you would be able to see more operations.
DataNucleus Core

Non-tx persist resulting in failure to persist can leave object in L1 cache as P_NEW

Created: 05/Jul/11 09:47 AM   Updated: 01/Aug/11 10:23 AM   Resolved: 05/Jul/11 11:13 AM
Component/s: Transactions
Affects Version/s: 3.0.0.m5
Fix Version/s: 3.0.0.m6


 Description  « Hide
When persisting non-transactional there is obviously no rollback. If we try to persist an object and it processes that object but during cascade persist it finds a problem the persist will fail but the first object is already in the L1 cache.

We likely need to
1. take a snapshot of the L1 cache before atomic non-tx persists and remove additional objects if there is a failure
2. disconnect any added StateManagers from these cached objects

Sort Order: Ascending order - Click to sort in descending order
Andy Jefferson added a comment - 05/Jul/11 11:13 AM
SVN trunk now fixes the behaviour in persistObject() so that any objects added to the L1 cache during the non-tx persist are removed if an exception is thrown. Doesn't do anything about disconnecting state managers yet since that will be much more work