Hi, We are using a 9 nodes cassandra cluster, and each node is currently handling about 40GB of data. Our CFs were created using LevedCompactionStrategy, with the default options (size of SSTables = 5MB). So the number of SSTables was going quite large, which led to problems described in CASSANDRA-4142 (OOM during repair).
Since then, we have updated our CFs definitions to set the sstable_size_in_mb option to 200MB. Now the newly created SSTables have the size we expect (about 200Mb). But now my question is : is there a way to compact or "scrub" the existing small SSTables so that they are re-created with the new size ? We didn't manage to do it with "nodetool scrub" : the sstables are recreated with their original size. Regards, Samuel