Does CRDB support backup/restore? Is the raft log format formalized and guaranteed to remain backward compatible? How can I dump raft logs and understand the underlying operations going on? I’m trying to explore integration points with other systems like Hadoop.
CRDB already supports a form of backup with
cockroach dump which produces a file that can be read with
cockroach sql. However these scale poorly with large amounts of data. We are working on an enterprise feature to allow for backup and restore to be done in parallel in a much faster way.
(Someone else will have to answer your questions about the raft log; I’m not familiar with them.)
You can inspect the raft logs of a stopped store with the
cockroach debug raft-log command, but raft logs are not exposed in any way from a running server. They’re full of implementation details and not what you’d want for integration with external systems anyway.