[CASSANDRA-18628] Handle special column name when upgrading from 2.2.19 to 3.0.29/3.11.15

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Not A Problem
• Component/s: CQL/Interpreter
• Platform: All
• Assignee: Unassigned
• Reporter: Ke Han
• Created: 26/Jun/23 21:27
• Updated: 26/Jun/23 22:07
• Resolved: 26/Jun/23 22:07

Description

When I was migrating from 2.2.19 to 3.0.29/3.11.15, I noticed some of my columns could not be accessed because of the syntax problem. I had a column named "Is" in the 2.x. However, after the upgrade process, when I try to perform the same select command, it throws an exception since "Is" will be interpreted separately. 

I observed this after my cluster was fully upgraded. I cannot downgrade it back to 2.2.19. I cannot rename this column directly either.

Steps to reproduce

Execute the following commands in Cassandra-2.2.19 (single node)

CREATE KEYSPACE  ks WITH REPLICATION = { 'class' : 'SimpleStrategy', 'replication_factor' : 2 };
CREATE TABLE IF NOT EXISTS ks.tb (c1 INT,c2 INT,c3 TEXT,rJr INT,c4 TEXT,c5 TEXT,ds INT, Is INT,
PRIMARY KEY (c2, c5, c3, rJr, Is, c1, ds, c4 ));

SELECT c2, ds, Is, c4 FROM ks.tb WHERE c2 = 1566089765 AND c5 = 'teststring1' AND c3 = 'teststring2' AND rJr = -1012111692; 

 c2 | ds | is | c4
----+----+----+----


(0 rows)

Then perform an upgrade to Cassandra-3.0.29/Cassandra-3.11.15

// Do the same select
SELECT c2, ds, Is, c4 FROM ks.tb WHERE c2 = 1566089765 AND c5 = 'teststring1' AND c3 = 'teststring2' AND rJr = -1012111692; 

SyntaxException: line 1:15 no viable alternative at input 'Is' (SELECT c2, ds, [Is]...)

The column still exists in the table, but I cannot access it.

I am wondering whether I can directly rename it in the new version or get some prompts telling me to rename is before the upgrade process. "Is" is also not marked as a reserved word in the CQL Keywords doc).

Any help is appreciated!