Missing client error and client disconnections with OLAP / streaming queries when using bad column names

[aquarapid]

TL;DR: If you use column names in streaming (workload=OLAP) queries, you can get strange results.

Setup:

• Current/recent master (56a3312)
• Setup local examples, unsharded commerce keyspace:

$ cd examples/local/
$ sh 101_initial_cluster.sh
.
.

• Connect to vtgate, and first see the behavior in normal (oltp) workload mode, and then in OLAP/streaming mode:

$ mysql -u root -P 15306 -h 127.0.0.1 -u root -A commerce
Welcome to the MySQL monitor.  Commands end with ; or \g.
Your MySQL connection id is 4
Server version: 5.7.9-Vitess MySQL Community Server - GPL

Copyright (c) 2000, 2020, Oracle and/or its affiliates. All rights reserved.

Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Other names may be trademarks of their respective
owners.

Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.

mysql> select * from product;
Empty set (0.01 sec)

mysql> select * from product order by price;
Empty set (0.00 sec)

mysql> select * from product order by col_does_not_exist;
ERROR 1054 (42S22): vtgate: http://localhost:15001/: target: commerce.0.master, used tablet: zone1-100 (localhost): vttablet: rpc error: code = NotFound desc = Unknown column 'col_does_not_exist' in 'order clause' (errno 1054) (sqlstate 42S22) (CallerID: userData1): Sql: "select * from product order by col_does_not_exist asc", BindVars: {}
mysql>
 
mysql> set workload=olap;
Query OK, 0 rows affected (0.00 sec)

mysql> select * from product order by price;
Empty set (0.00 sec)

mysql> select * from product order by col_does_not_exist;
Query OK, 0 rows affected (0.00 sec)                                           <----  This should not happen

mysql> select * from product order by price;
ERROR 2013 (HY000): Lost connection to MySQL server during query
mysql> select * from product order by price;
ERROR 2006 (HY000): MySQL server has gone away
No connection. Trying to reconnect...
Connection id:    5
Current database: commerce

Empty set (0.00 sec)

Note the "Lost connection". If you look at the vtgate logs, it becomes evident what happened:

• During the select * from product order by col_does_not_exist; query in OLAP mode, vtgate reports the error:

E0912 14:36:20.284310   12811 conn.go:1184] Error in the middle of a stream to client 4 (127.0.0.1:53098): vtgate: http://localhost:15001/: target: commerce.0.master, used tablet: zone1-100 (localhost): vttablet: rpc error: code = NotFound desc = Unknown column 'col_does_not_exist' in 'order clause' (errno 1054) (sqlstate 42S22) (CallerID: userData1): Sql: "select * from product order by col_does_not_exist asc", BindVars: {} (errno 1054) (sqlstate 42S22)

• However, this error does not make it to the client, but only on the next query does it become evident that there was an issue (Lost connection)

[aquarapid]

Should also note that the same error occurs in OLAP/streaming mode when you use a column that does not exist in the SELECT clause, instead of the ORDER BY, e.g.: select col_does_not_exist from product;

[aquarapid]

I believe this was fixed by the fix for #6934