-
Bug
-
Resolution: Unresolved
-
Major
-
None
-
7.2.2.Final
-
None
I observed this in the server log of a WildFly CI run (https://ci.wildfly.org/viewLog.html?buildId=523078&buildTypeId=WF_PullRequest_LinuxSmJdk17&tab=buildResultsDiv)
Caused by: java.lang.IllegalStateException: Can't overwrite cause with org.omg.CosTransactions.InvalidControl: IDL:omg.org/CosTransactions/InvalidControl:1.0 at java.base/java.lang.Throwable.initCause(Throwable.java:463) at org.jboss.jts@7.2.2.Final//com.arjuna.ats.internal.jta.transaction.jts.TransactionManagerImple.resume(TransactionManagerImple.java:153) at org.jboss.jts.integration@7.2.2.Final//com.arjuna.ats.jbossatx.BaseTransactionManagerDelegate.resume(BaseTransactionManagerDelegate.java:113) at org.wildfly.transaction.client@3.0.5.Final//org.wildfly.transaction.client.LocalTransaction.resume(LocalTransaction.java:253) at org.wildfly.transaction.client@3.0.5.Final//org.wildfly.transaction.client.ContextTransactionManager.resume(ContextTransactionManager.java:158) at org.wildfly.transaction.client@3.0.5.Final//org.wildfly.transaction.client.ContextTransactionManager.resume(ContextTransactionManager.java:150) at org.jboss.as.ejb3@38.0.0.Final-SNAPSHOT//org.jboss.as.ejb3.tx.CMTTxInterceptor.safeResume(CMTTxInterceptor.java:303) ... 73 more Caused by: jakarta.transaction.InvalidTransactionException at org.jboss.jts@7.2.2.Final//com.arjuna.ats.internal.jta.transaction.jts.TransactionManagerImple.resume(TransactionManagerImple.java:152) ... 78 more
The problematic code is this in TransactionManagerImple:
InvalidTransactionException invalidTransactionException = new InvalidTransactionException();
invalidTransactionException.initCause(e1);
InvalidTransactionException is a subclass of RemoteException, which explicitly disables the ability to call initCause by setting the cause (to null) in its own constructor:
public RemoteException() {
initCause(null); // Disallow subsequent initCause
}
Unfortunately, invalidTransactionException doesn't expose a constructor that allows passing in the cause that way. (At least not in this version; IIRC Jakarta Transactions recently changed that for EE 12.)
Note: I don't know if this has anything to do with the linked test failing. I doubt it as this looks like a problem in dealing with an error condition the test is not trying to create. So the original error condition would likely have caused some problem on its own.
