Closed charvolant closed 6 years ago
Ive added this for cassandra 3. Just a single file based checkpoint thing which uses the index token ranges. Works well for processing and indexing. For indexing, theres an additional step that deletes entries from a partially indexing token range before re-starting indexing.
this is done for version 2.x and i dont think we'll do for 1.9.
It would be helpful to have checkpointing for bulk processing and bulk indexing operations, so that if the job falls over halfway through, it can start from a last known good point, rather than from the beginning. That way, if a 4-day job falls over in the last hour, it can be restarted easily.