Details
-
Improvement
-
Status: Resolved
-
Low
-
Resolution: Fixed
-
None
Description
The situation encountered is that after deleting the columns for a row in the cli new workloads don't update that columns for the rows. Brandon mentioned that stress uses millis, cli uses micros and therefore row tombstone wins.
Test Case: Before Delete
// Run: stress --operation=INSERT --num-keys=1 --columns=2 --consistency-level=QUORUM --column-size=1 --threads=1 --replication-factor=1 --nodes=localhost // Run cassandra-cli [default@Keyspace1] list Standard1; Using default limit of 100 ------------------- RowKey: 30 => (column=C0, value=63, timestamp=1326259090065) => (column=C1, value=63, timestamp=1326259090065) 1 Row Returned. Elapsed time: 2 msec(s). [default@Keyspace1] del Standard1['30']; row removed. [default@Keyspace1] list Standard1; Using default limit of 100 ------------------- RowKey: 30
Test Case: After Delete
// Run: stress --operation=INSERT --num-keys=1 --columns=2 --consistency-level=QUORUM --column-size=1 --threads=1 --replication-factor=1 --nodes=localhost // Run cassandra-cli [default@Keyspace1] list Standard1; Using default limit of 100 ------------------- RowKey: 30 1 Row Returned. Elapsed time: 1 msec(s).