Cassandra has always supported two great ways to prevent data loss: * Replication * Backups
I doubt Cassandra will ever focus extensively on single-node recovery when it's so easy to wipe and rebuild any node from the cluster. -----Original Message----- From: JKnight JKnight <beukni...@gmail.com> Date: Wed, 31 Mar 2010 03:48:01 To: <cassandra-u...@incubator.apache.org> Subject: Cassandra data file corrupt Dear all, My Cassandra data file had problem and I can not get data from this file. And all row after error row can not be accessed. So I lost a lot of data. Will next version of Cassandra implement the way to prevent data lost. Maybe we use the checkpoint. If data file corrupt, we will read from the next checkpoint. If not, can you suggest me the way to implement this function? -- Best regards, JKnight
