[jira] [Commented] (AIRFLOW-4464) Fix case-insensitive id columns in mysql

["ASF GitHub Bot (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/AIRFLOW-4464?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16927665#comment-16927665 ] 

ASF GitHub Bot commented on AIRFLOW-4464:
-----------------------------------------

stale[bot] commented on pull request #5244: [AIRFLOW-4464] Add regression test for case-insensitive ids in mysql.
URL: https://github.com/apache/airflow/pull/5244
 
 
   
 
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


> Fix case-insensitive id columns in mysql
> ----------------------------------------
>
>                 Key: AIRFLOW-4464
>                 URL: https://issues.apache.org/jira/browse/AIRFLOW-4464
>             Project: Apache Airflow
>          Issue Type: Improvement
>          Components: database
>            Reporter: Josh Carp
>            Assignee: Josh Carp
>            Priority: Minor
>              Labels: mysql
>
> By default, string comparisons in mysql are case-insensitive, so the task ids "foo" and "FOO" are treated as identical. This means that a dag with those task ids will fail to schedule with a sqlalchemy `IntegrityError` using mysql, but not postgres or sqlite. This situation probably doesn't happen often, and users probably shouldn't use task ids that are identical except for case, but I think we should improve the behavior here. A few options:
>  
>  * Configure sqlalchemy to use a binary collation for string id columns under mysql so that string comparisons are case-sensitive.
>  * Require dag and task ids to be unique regardless of case. This would be a breaking change.
>  * Document that mysql users should configure mysql to use binary collations for string types by default. This would still show users a 500 if the database isn't configured correctly.
>  
> I'll submit a pull request with a failing unit test to describe the issue.



--
This message was sent by Atlassian Jira
(v8.3.2#803003)