After upgrading to KNIME Analytics Platform 4 (currently 4.0.2) I started experiencing an AWT deadlock that is reported to the log with thread dump following it.
2019-11-26 13:58:10,236 DEBUG MemoryAlertSystem Memory threshold of 87% exceeded
2019-11-26 13:58:10,251 DEBUG MemoryAlertSystem 0/0 listeners have been removed, 0 are remaining
2019-11-26 13:59:11,852 DEBUG MemoryAlertSystem Estimated tenured gen heap space usage below threshold (87%), currently 78% (3.11GB/4.00GB)
2019-11-26 14:01:55,860 DEBUG MemoryAlertSystem Memory threshold of 87% exceeded
2019-11-26 14:01:55,930 DEBUG MemoryAlertSystem 0/0 listeners have been removed, 0 are remaining
2019-11-26 14:01:56,219 DEBUG KNIMEApplication$4 Potential deadlock in AWT Event Queue detected. Full thread dump will follow as debug output.
2019-11-26 14:01:56,219 DEBUG KNIMEApplication$4 onditionObject@59d93b26
This usually occurs while a workflow is processing larger tables of about a few million on lower tens of million records. Please note I didn’t configure my nodes with “keep tables in memory” option and I even configured my KNIME with -Dknime.table.cache=SMALL added to the knime.ini file. Xmx option is set to 4G.
I run my workflows on MacBook with 16GB RAM but I’ve experienced the same behaviour on Windows 10 laptop with 8GB RAM too. KNIME consumes about 5,5GB RAM while running the workflow. IDE is responsive on Mac whereas often or fully irresponsive on Windows.
I’m attaching related part of log.
KNIME_AWT_EQ_DeadLock.log (174.6 KB)