Upgrading MySQL charset to utf8mb4

[dave (Dave Ormsbee)]

@Abdess

Only noticed one strange thing during the migration. The title of one of the courses “Apprendre à apprendre” became “Apprendre à apprendre”. And that’s only for the title of the course, everything else is well written.

That is a little troubling. I’d be curious to know what the exported data looks like for that entry.

@iamCristYe

If my understanding is correct, to avoid downtime, we can do the following:

1. update the default database charset. (This is used only as a default value for future tables)
2. update the default table charset (This is used only as a default value for future columns, at this point utf8mb4 still cannot be inserted)
3. update the column charset (at this point utf8mb4 can be inserted via MySQL CLI but not through Django as it’s still using utf8mb3 connection?)
4. redeploy the Django application with the new utf8mb4 connection settings

Ah, maybe that’s the part I’m missing. I didn’t realize that it was possible to incrementally upgrade individual tables to utf8mb4 while the Django connection is utf8 and still have everything work normally for the 3-byte UTF-8 characters. If that’s the case, that’s fantastic, particularly for large sites that need to be extra careful about performance regressions.