I have a MySQL table that has the collation for one of its varchar columns in the primary key set to utf8_general_ci. I want to change the collation to utf8_unicode_ci so it stops treating "gmail.com" and "gmàil.com" like they're the same item. I tried running this:

alter table my_table change character set utf8 collate utf8_unicode_ci;

The alter fails with a duplicate key violation on one of the items.

Error 1062 (23000): Duplicate entry 2013-07-15-119-comcast.net for key PRIMARY

I tried selecting the items out of the table that have date equal to 2013-07-15 and source_id equal to 119, but eyeballing the results I don't see two entries for "comcast.net" that might be causing the problem.

Can I construct a query into this table to determine which two rows are different under utf8_general_ci but the same under utf8_unicode_ci?

  • best solution - convert the table to the same collation with all other, many benefits and prevent other collisions
    – a_vlad
    Mar 6, 2016 at 20:40

1 Answer 1


A query like this did the trick:

select date, thing_id, email_domain collate utf8_unicode_ci
from my_table
where date = '2013-07-15'
and thing_id = 119
and email_domain collate utf8_unicode_ci = 'comcast.net';

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.