I've cloned one cluster to another with these steps:

  1. Copy table files on each source node from /var/lib/cassandra/data/keyspace/table/snapshots... to /location/snaps/

  2. Run sstableloader on each source node.

  3. Run nodetool import keyspace table /var/lib/cassandra... on each destination node.

After these steps I get this error when I try to run:

select count (*) from keyspace.table; 

ReadFailure: Error from server: code=1300 [Replica(s) failed to execute read] message="Operation failed - received 0 responses and 1 failures (UNKNOWN=[/])" info={'consistency': 'ONE', 'required_responses': 1, 'received_responses': 0, 'failures': 1, 'error_code_map': {'': '0x0000'}}

Nodetool repair doesn't make any difference.

1 Answer 1


A ReadFailure error is returned by the request coordinator when one of the replicas replies with a read error.

In my experience, this is almost always caused by a read request reaching the tombstone_failure_threshold (there were more than 100,000 tombstones scanned). You will need to check the logs on the offending node to confirm it or it will otherwise provide clues for the failure.

As a side note, COUNT() doesn't scale in a distributed architecture and you're likely to run into a timeout because Cassandra took too long to count the records across all nodes. I've discussed it in detail in Why is COUNT() bad in Cassandra?.

The DataStax Bulk Loader (DSBulk) tool has a count command which was written specifically for this purpose. DSBulk's count command overcomes the problems with the CQL COUNT() function by dividing the token range into small segments and performing a distributed count.

Here are some references with examples to help you get started quickly:

DSBulk is fully open-source so it's free to use. Cheers!

  • I've tried select * from keyspace.table limit 10; and it returned 10 rows correctly. Thanks a lot! Commented Mar 2, 2023 at 7:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.