Update the dev environment to the current prod database schema #2268
Conversation
There was a problem hiding this comment.
Please normalize the form of the sql file so that it is the same as produced by npm run publish:db on the dev docker container. That way future change sets will show only actual changes to the database. (To do this, execute npm run reset:db followed by npm run publish:db. Attempting this allowed me to catch the other error.)
Also, if we're changing db_template.sql, I suggest reviewing all the INSERT queries that initialize data that were present before this commit. I'm not sure that we necessarily want to initialize the way we do. Maybe we can discuss if there a reason for those initial records. Some don't seem to make sense. One option is to remove most of the INSERT queries.
| CONSTRAINT `course_user_ibfk_1` FOREIGN KEY (`roleId`) REFERENCES `course_role` (`roleId`) ON UPDATE CASCADE | ||
| ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb3 COLLATE=utf8mb3_unicode_ci; | ||
| KEY `driveId_userId` (`courseId`,`userId`) | ||
| ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci; |
There was a problem hiding this comment.
I can't add a comment on directly on line 429, below, so I'm adding it here. The INSERT statement on line 429 is now invalid given that the number of columns does not match. Error when try npm run reset:db is:
ERROR 1136 (21S01) at line 429: Column count doesn't match value count at row 1
I'll do the suggested normalization and fix the insert that now doesn't match the schema. I agree that we should review the insert queries. To generate this diff I had to do a merge resolution with the thing I dumped out of prod, which was schema-only, and the db_template currently checked into main, which had these insertion records. It would make the process of auditing prod against dev easier in the future if we separated any desired default record insertion from the schema. |
No description provided.