hibernate/hibernate-orm · error · TransactionException

Unable to resume suspended transaction

Error message

Unable to resume suspended transaction

What it means

TransactionException from JtaIsolationDelegate's finally block: the isolated work succeeded, but resuming the suspended JTA transaction failed. (When the work itself failed, the resume error is attached as a suppressed exception instead — this standalone throw means the work is committed/done while the surrounding transaction context is lost.) The caller's transaction must be abandoned and restarted.

Source

Thrown at hibernate-core/src/main/java/org/hibernate/resource/transaction/backend/jta/internal/JtaIsolationDelegate.java:129

			return callable.call();
		}
		catch ( HibernateException he ) {
			exception = he;
			throw he;
		}
		catch ( Throwable throwable ) {
			exception = throwable;
			throw new HibernateException( "Unable to perform isolated work", throwable );
		}
		finally {
			try {
				// resume the JTA transaction we suspended
				resume( surroundingTransaction );
			}
			catch ( Throwable throwable ) {
				// if the actual work had an error, use that; otherwise throw this error
				if ( exception == null ) {
					throw new TransactionException( "Unable to resume suspended transaction", throwable );
				}
				else {
					exception.addSuppressed( throwable );
				}
			}
		}
	}

	private void resume(Transaction surroundingTransaction) throws InvalidTransactionException, SystemException {
		if ( surroundingTransaction != null ) {
			transactionManager.resume( surroundingTransaction );
			JTA_LOGGER.transactionResumed( surroundingTransaction );
		}
	}

	private Transaction suspend() throws SystemException {
		final var surroundingTransaction = transactionManager.suspend();
		if ( surroundingTransaction != null ) {

View on GitHub (pinned to fad1729dce)

Solutions

1. Abort and restart the outer transaction — it cannot be safely continued after a failed resume
2. Increase the JTA transaction timeout above worst-case isolated-work duration
3. Move long isolated work (DDL, big generator segment allocation) out of transactional scope
4. Inspect TM logs for the resume failure reason before retrying

Example fix

# before: default 60s timeout, isolated work inside a batch job exceeds it
# WildFly: /subsystem=transactions:write-attribute(name=default-timeout,value=60)

# after: raise the timeout for jobs that suspend/resume transactions
# WildFly: /subsystem=transactions:write-attribute(name=default-timeout,value=300)

catch (org.hibernate.TransactionException e) {
  // resume failed after successful isolated work: treat the outer transaction as lost —
  // roll back / abandon it, verify what the isolated work did on the DB, restart fresh
}

When it happens

Trigger: After isolated work (ID generation, DDL helper) under JTA, TransactionManager.resume() throws — the suspended transaction timed out or was rolled back while the isolated work ran, or the TM is failing mid-operation.

Common situations: JTA transaction timeouts shorter than the duration of isolated work inside batch jobs; TM recovery or failover during long operations; heavy load extending work past the default timeout.

Related errors

• Unable to resume JTA transaction after DDL execution
• Unable to apply requested transaction timeout
• Unable to apply requested transaction timeout
• Named query hint [" + hintName + "] is not an integer: " + q
• Could not obtain TransactionManager from JtaPlatform

AI-assisted analysis of hibernate/hibernate-orm@fad1729dce (2026-08-22). Data as JSON: /api/errors/683f8507f5326792. Report an issue: GitHub.