Hi Alberto, Can you share the full query, configured indexes and configs? Both JVM and Ignite.
Thanks, Evgenii пт, 5 окт. 2018 г. в 22:00, Alberto Mancini <ab.manc...@gmail.com>: > Hi, > we are developing a poc using Ignite and we faced a scaring problem > that i suppose is due to some naive misconfiguration. > > Long story short, we use persistence on a cluster on AWS and to speedup > test data loading we attempted a simple sql query: > > INSERT INTO MyTable ( > ID > PARTITION_KEY, > ... > ) > SELECT > ID, > 'another_partition_key', > ... > ) > > With enough rows this end up killing nodes. > With the default failureDetectionTimeout the server nodes > end up offline, tuning the timeout we get OOM. > > Obviously the query is used only for testing but the fact that the server > nodes stop is actually scaring. > > Do we missed some obvious configuration that may keep healthy the cluster? > > Thanks, > A. >