Reputation: 13
After scanning 100k tombstones, cassandra will error out the query, I tried to do major compaction on the table, but it doesn't remove tombstones of its secondary index. The query still can not be completed.
I searched for a while, one suggestion is rebuild_index, but I think it will cause many queries to fail while rebuilding, and I don't have an estimate of how long it'll take to rebuild indexes.
Any suggestion?
Upvotes: 0
Views: 1002
Reputation: 53
I don't know if you found your answer but I hope it can help anyone : Try to rebuild to secondary index, I suspect that ordinary nodetool command does not "clean" seconday index properly.
# nodetool rebuild_index keyspace "CF" "CF".SecondayIndex
Like Ananth said, you might have to rethink your datamodel to avoid secondary index being full with tombstones again.
Regards,
Upvotes: 3
Reputation: 991
Change the gc_grace_seconds to 1 and run a compaction on the table and see. It should clear the tombstones immediately. Once done,reset the gc_grace_seconds to a lower value. Default is 10 days. I think you need to model your data better.
Upvotes: -1