From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040204 Galeon/1.3.12 Description of problem: The Java code does not always ensure that we have exactly one transaction per HTTP request. In some cases we commit more than once during a request, making it possible to leave objects in an inconsistent state that the user has no way of fixing. UserManager.createUser is an example of this. See the thread at http://post-office.corp.redhat.com/archives/rhn-java-list/2005-April/msg00169.html for more discussion Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: Review code Additional info:
Committed a TransactionFactory at r53746
Fixed estimate and updated with progress
Txn handling is now done in SessionFilter. Nesting txns will produce errors. r54527
Update effort info
mass move to ON_QA
testplan?
This bug is not testable by QA. The only indication that something go wrong is that if a user performs a UI action that changes something, and that action fails with a server error, changes may have been written to teh database. We have unit tests to verify that this does not happen; Bret: what's the best proper way to close out this bug ?
mass move for 7/14 qa push. picking up jesusr's and shughes' since they're both out, as well.
Setting rnorwood as qa contact for java migration bugs
Moving this to PROD_READY, since as david says, there's not a good way to QA this directly.