We are running ignite 2.0.0 with the Arjuna transaction manger, configuration snippet follows:
igniteConfiguration.getTransactionConfiguration().setTxManagerFactory((Factory<TransactionManager>) () -> jtaTransactionManager); igniteConfiguration.getTransactionConfiguration().setUseJtaSynchronization(true); config.setWriteThrough(true); config.setCacheMode(CacheMode.REPLICATED); config.setAtomicityMode(CacheAtomicityMode.TRANSACTIONAL); This seems to work well and it appears that a rollback triggered by other parts of the business logic cause cache changes to be rolled back. But then someone did OS level security patches on the database server where ignite writes its cache data. While Arjuna rolls back the transactions happening while the DB is down, it appears ignite goes into a downward spiral of "GridCachePartitionExchangeManager - Found long running transaction". These never go away. We would think this should have been killed by Arjuna as it rolls back ? We did not set setDefaultTxTimeout in transactionConfiguration, since we think Arjuna should handle it. What am I missing here ? Kristian