Question

Why does simply changing the collation cause a duplicate key error?

mysql> describe phppos_items;
+-----------------------+--------------+------+-----+---------+----------------+
| Field                 | Type         | Null | Key | Default | Extra          |
+-----------------------+--------------+------+-----+---------+----------------+
| name                  | varchar(255) | NO   | MUL | NULL    |                |
| category              | varchar(255) | NO   | MUL | NULL    |                |
| supplier_id           | int(11)      | YES  | MUL | NULL    |                |
| item_number           | varchar(255) | YES  | UNI | NULL    |                |
| description           | varchar(255) | NO   |     | NULL    |                |
| cost_price            | double(15,2) | NO   |     | NULL    |                |
| unit_price            | double(15,2) | NO   |     | NULL    |                |
| quantity              | double(15,2) | NO   |     | 0.00    |                |
| reorder_level         | double(15,2) | NO   |     | 0.00    |                |
| location              | varchar(255) | NO   |     | NULL    |                |
| item_id               | int(10)      | NO   | PRI | NULL    | auto_increment |
| allow_alt_description | tinyint(1)   | NO   |     | NULL    |                |
| is_serialized         | tinyint(1)   | NO   |     | NULL    |                |
| deleted               | int(1)       | NO   | MUL | 0       |                |
+-----------------------+--------------+------+-----+---------+----------------+
14 rows in set (0.01 sec)

mysql> ALTER TABLE  `phppos_items` CHANGE  `name`  `name` VARCHAR( 255 ) CHARACTER SET utf8 COLLATE utf8_unicode_ci NOT NULL , CHANGE  `category`  `category` VARCHAR( 255 ) CHARACTER SET utf8 COLLATE utf8_unicode_ci NOT NULL , CHANGE  `item_number`  `item_number` VARCHAR( 255 ) CHARACTER SET utf8 COLLATE utf8_unicode_ci NULL DEFAULT NULL , CHANGE  `description`  `description` VARCHAR( 255 ) CHARACTER SET utf8 COLLATE utf8_unicode_ci NOT NULL , CHANGE  `location`  `location` VARCHAR( 255 ) CHARACTER SET utf8 COLLATE utf8_unicode_ci NOT NULL;
ERROR 1062 (23000): Duplicate entry ' ' for key 'item_number'

Create table:

| phppos_items | CREATE TABLE `phppos_items` (
  `name` varchar(255) CHARACTER SET latin1 NOT NULL,
  `category` varchar(255) CHARACTER SET latin1 NOT NULL,
  `supplier_id` int(11) DEFAULT NULL,
  `item_number` varchar(255) CHARACTER SET latin1 DEFAULT NULL,
  `description` varchar(255) CHARACTER SET latin1 NOT NULL,
  `cost_price` double(15,2) NOT NULL,
  `unit_price` double(15,2) NOT NULL,
  `quantity` double(15,2) NOT NULL DEFAULT '0.00',
  `reorder_level` double(15,2) NOT NULL DEFAULT '0.00',
  `location` varchar(255) CHARACTER SET latin1 NOT NULL,
  `item_id` int(10) NOT NULL AUTO_INCREMENT,
  `allow_alt_description` tinyint(1) NOT NULL,
  `is_serialized` tinyint(1) NOT NULL,
  `deleted` int(1) NOT NULL DEFAULT '0',
  PRIMARY KEY (`item_id`),
  UNIQUE KEY `item_number` (`item_number`),
  KEY `phppos_items_ibfk_1` (`supplier_id`),
  KEY `name` (`name`),
  KEY `category` (`category`),
  KEY `deleted` (`deleted`),
  CONSTRAINT `phppos_items_ibfk_1` FOREIGN KEY (`supplier_id`) REFERENCES `phppos_suppliers` (`person_id`)
) ENGINE=InnoDB AUTO_INCREMENT=1560 DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci |

Answer 1

When modifying the field collation, perhaps MySQL is as well trying to convert the data from latin1 to utf8. It might be that after this conversion, some of your data in the item_number column is containing duplicates, a space character perhaps it seems from the error.

I think these steps might help to identify which are the conflicting rows:

1. select item_id, name and item_number columns for all rows and keep the entries in some XLS file perhaps
2. remove the unique index on item_number column
3. run the ALTER TABLE statement that should now run successfully
4. redo step 1 and compare the output with that from step 1 and find out for which of the items is item_number bearing the same values

Hope this helps!