Running dsc21 cassandra 2.1.5 and has this in our cassandra.yaml:
compaction_large_partition_warning_threshold_mb: 512000
but are still see warnings like these:
WARN [MemtableFlushWriter:1] 2017-01-08 12:11:11,626 SSTableWriter.java:241 - Compacting large partition system/local:local (472 bytes) WARN [MemtableFlushWriter:2] 2017-01-08 12:11:11,688 SSTableWriter.java:241 - Compacting large partition system/local:local (126 bytes) WARN [CompactionExecutor:1] 2017-01-08 12:11:11,804 SSTableWriter.java:241 - Compacting large partition system/local:local (12567 bytes)
thus sees tons of warnings getting logged :/
Why still log when sizes are way below requested threshold, bug or feature?
Been told that internally this config settinbg is multiplied w/1M and stored in a java int data type of a signed 32bit => 2^31 ~ 2048M, thus setting threshold to 2000M stopped the excessive logging :)