Since upgrading to 6.0.1 - which included a move of the CLDB to a good system - we have been receiving CLDB memory usage alarms with this notification:
More than 10 buckets of size <= 8388608 have been flushed in last 300000 ms, increase memory or add nodes as this will cause performance degradation.
Searching for this seems to suggest this is related to inserts into MapR-DB being made by a StreamSets stream - which is also new. The CLDB in question has 256 GB RAM (The old one had only 48GB) so it seems like this is probably not the issue. I'm not sure where or what processes to "tune" to fix the problem - is this a CLDB or a MapR-DB issue?
Thanks for any insight or suggestions.