Kill the connection on the specified thread. Rolls back any query running on the thread.


KILL CONNECTION <internal_process_id> [<node_id>]
KILL QUERY <query_id> [<node_id>]


  • internal_process_id - ID of the thread to kill.

  • query_id - ID of the query to kill.

  • node_id - ID of the node (aggregator) on which the query/thread is running. If the node_id is not specified, the query/thread running on the current node is killed.


    You can query the information_schema.MV_PROCESSLIST table for the required parameter values, where the ID field in the table contains the thread/query ID and the NODE_ID field contains the ID of the node.


  • internal_process_id - ID of the thread to kill, which can be found by running SHOW_PROCESSLIST.

  • KILL CONNECTION will disconnect the client associated with internal_process_id. This is the default, meaning that running KILL <> will have the same result.

  • KILL QUERY allows you to kill a query running on any aggregator in the cluster.

  • KILL QUERY will not disconnect the client associated with internal_process_id. Instead, the client will receive an error.

  • Any query running on the connection will be rolled back.

  • SingleStoreDB checks for the kill bit during any potentially long running loop during query execution and rolls back the query’s transaction if the kill bit is set.

  • This command should be run on the master aggregator or a child aggregator node (see Node Requirements for SingleStoreDB Commands).

  • Refer to the Permission Matrix for the required permission.


Query OK, 0 rows affected (0.00 sec)

Related Topics

Last modified: January 9, 2023

Was this article helpful?