On Fri, Dec 27, 2013 at 6:13 PM, Josh Dzielak <j...@keen.io> wrote: > Our suspicion is that we somehow have a row level tombstone that > is future-dated and has not gone away (we’ve lowered gc_grace_seconds in > hope that it’d get compacted, but no luck so far, even though the stables > that hold the row key have all cycled since). >
What version of Cassandra? ecapriolo is right, use sstablekeys/sstable2json to inspect suspect rows in SSTables. If that's the problem, you can follow this procedure : http://thelastpickle.com/2011/12/15/Anatomy-of-a-Cassandra-Partition/ Or you can dump/reload with sstable2json/json2sstable and filter out bad values. =Rob