[CASSANDRA-16372] Import from csv of empty strings in list fails with a ParseError: Empty values are not allowed, given up without retries

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 3.0.24, 3.11.10, 4.0-rc1, 4.0
• Component/s: Tool/cqlsh
• Bug Category: Correctness
• Severity: Normal
• Complexity: Normal
• Discovered By: User Report
• Platform: All
• Since Version: 3.0.0

Source Control Link

https://github.com/apache/cassandra/commit/a542b865047ad8cc3d9f64ab3c3a018ac069331e

Test and Documentation Plan

dtests patch adds test cases with empty strings inside collections. Already existing test_reading_collections_with_empty_values covers the case where empty values for other data types still throw during parsing rather than on the server-side.

• Assignee: Alex Sorokoumov
• Reporter: Ostico

Authors

Alex Sorokoumov

Reviewers

Brandon Williams

• Created: 04/Jan/21 15:23
• Updated: 13/Jan/21 09:47
• Resolved: 12/Jan/21 19:32

Description

 

Cqlsh fail to import an empty string which is present in a list data type.

In those conditions, simple csv backups can discard rows and data can be corrupted.

 

Conditions

 

# cqlsh
Connected to Test Cluster at 127.0.0.1:9042.
[cqlsh 5.0.1 | Cassandra 3.11.6 | CQL spec 3.4.4 | Native protocol v4]
Use HELP for help.

CREATE TABLE test.test_1 (
    uid uuid PRIMARY KEY,
    texts list<text>
);

insert into test.test_1 ( uid, texts ) VALUES( 833fee3f-d4f9-418b-9387-84ac2cda5cb7, ['But if you now try to wash your hands,', ''] );

 

Now exporting and and re-importing data fails:

 

cqlsh> select * from test.test_1;

 uid                                  | texts
--------------------------------------+------------------------------------------------
 833fee3f-d4f9-418b-9387-84ac2cda5cb7 | ['But if you now try to wash your hands,', '']


cqlsh> COPY test.test_1 (uid, texts) TO 'ctm.csv'; 
Using 7 child processesStarting copy of test.test_1 with columns [uid, texts].
Processed: 1 rows; Rate: 9 rows/s; Avg. rate: 9 rows/s 1 rows exported to 1 files in 0.148 seconds. 

cqlsh> truncate table test.test_1;


cqlsh> COPY test.test_1 (uid, texts) FROM 'ctm.csv';
Using 7 child processes
Starting copy of test.test_1 with columns [uid, texts].
Failed to import 1 rows: ParseError - Failed to parse ['But if you now try to wash your hands,', ''] : Empty values are not allowed,  given up without retries
Failed to process 1 rows; failed rows written to import_test_test_1.err
Processed: 1 rows; Rate:       2 rows/s; Avg. rate:       2 rows/s
1 rows imported from 1 files in 0.415 seconds (0 skipped).


cqlsh> select * from test.test_1; 
uid  | pid
-----+-----