I've got a bizarre issue with MySQL (specifically MySQL 5.5.5-10.0.26-MariaDB) where it is throwing the "Row size too large" error when (by my logic!) it shouldn't be.
I have an ExpressionEngine installation that was recently migrated to InnoDB from MyISAM for performance reasons. The main entry table is throwing errors when trying to update large entries:
> Row size too large (> 8126). Changing some columns to TEXT or BLOB or using ROW_FORMAT=DYNAMIC or ROW_FORMAT=COMPRESSED may help. In current row format, BLOB prefix of 768 bytes is stored inline.
The `UPDATE` query ExpressionEngine is trying to run is huge (updating around 50 columns, with a lot of large content). But this error is only ever thrown when updating one specific `TEXT` column (`field_id_13`), with data only 1500 bytes in length.
- If I comment out the setting of `field_id_13`, the query works every single time
- If I comment out `field_id_13` and add extra fields totalling more data being inserted, the query still works!
- It's only this one column that is causing the error - commenting or adding any other columns gives no problems at all
The column is being set with the following data:
`field_id_13` = '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',
That's 1540 characters of data. The data for `field_id_13` is base64-encoded data of a serialised PHP object (not my design decision unfortunately - this is done by an EE field type). When decoded, the string length is only 1154 characters.
The column `field_id_13` is set as `TEXT`, though I have also tried setting it as `VARCHAR(10000)`, `MEDIUMTEXT`, `LONGTEXT`, and `BLOB` - neither of which make any difference as the error is always thrown.
The MySQL installation is using the Antelope file system. I am aware a move can be made to Barracuda and I can use `ROW_FORMAT=compressed` but the client is very performance-conscious and I am reluctant to bring the increase in disk usage overhead for compressing/decompressing on the fly.
Any ideas what's going on?
For reference, here is my `CREATE TABLE` result:
CREATE TABLE `exp_channel_data` (
`entry_id` int(10) unsigned NOT NULL,
`site_id` int(4) unsigned NOT NULL DEFAULT '1',
`channel_id` int(4) unsigned NOT NULL,
`field_id_2` text,
`field_ft_2` tinytext,
`field_id_3` text,
`field_ft_3` tinytext,
`field_id_4` text,
`field_ft_4` tinytext,
`field_id_5` text,
`field_ft_5` tinytext,
`field_id_12` mediumtext,
`field_ft_12` tinytext,
`field_id_13` text,
`field_ft_13` tinytext,
`field_id_18` mediumtext,
`field_ft_18` tinytext,
`field_id_19` mediumtext,
`field_ft_19` tinytext,
`field_id_20` text,
`field_ft_20` tinytext,
`field_id_21` mediumtext,
`field_ft_21` tinytext,
`field_id_22` mediumtext,
`field_ft_22` tinytext,
`field_id_23` text,
`field_ft_23` tinytext,
`field_id_24` int(11) DEFAULT '0',
`field_ft_24` tinytext,
`field_id_25` mediumtext,
`field_ft_25` tinytext,
`field_id_26` text,
`field_ft_26` tinytext,
`field_id_46` float DEFAULT '0',
`field_ft_46` tinytext,
`field_id_47` mediumtext,
`field_ft_47` tinytext,
`field_id_49` text,
`field_ft_49` tinytext,
`field_id_50` text,
`field_ft_50` tinytext,
`field_id_51` text,
`field_ft_51` tinytext,
`field_id_52` text,
`field_ft_52` tinytext,
`field_id_53` mediumtext,
`field_ft_53` tinytext,
`field_id_54` text,
`field_ft_54` tinytext,
`field_id_55` text,
`field_ft_55` tinytext,
`field_id_56` text,
`field_ft_56` tinytext,
`field_id_91` text,
`field_ft_91` tinytext,
`field_id_92` text,
`field_ft_92` tinytext,
`field_id_93` text,
`field_ft_93` tinytext,
`field_id_94` text,
`field_ft_94` tinytext,
`field_id_95` text,
`field_ft_95` tinytext,
`field_id_96` text,
`field_ft_96` tinytext,
`field_id_97` text,
`field_ft_97` tinytext,
`field_id_98` text,
`field_ft_98` tinytext,
`field_id_99` text,
`field_ft_99` tinytext,
`field_id_100` text,
`field_ft_100` tinytext,
`field_id_101` text,
`field_ft_101` tinytext,
`field_id_102` text,
`field_ft_102` tinytext,
`field_id_103` text,
`field_ft_103` tinytext,
`field_id_104` text,
`field_ft_104` tinytext,
`field_id_105` text,
`field_ft_105` tinytext,
`field_id_106` text,
`field_ft_106` tinytext,
`field_id_107` mediumtext,
`field_ft_107` tinytext,
`field_id_109` text,
`field_ft_109` tinytext,
`field_id_110` text,
`field_ft_110` tinytext,
`field_id_111` text,
`field_ft_111` tinytext,
`field_id_112` text,
`field_ft_112` tinytext,
`field_id_115` text,
`field_ft_115` tinytext,
`field_id_116` text,
`field_ft_116` tinytext,
`field_id_119` text,
`field_ft_119` tinytext,
`field_id_120` text,
`field_ft_120` tinytext,
`field_id_121` text,
`field_ft_121` tinytext,
`field_id_122` text,
`field_ft_122` tinytext,
`field_id_123` text,
`field_ft_123` tinytext,
`field_id_124` text,
`field_ft_124` tinytext,
`field_id_125` text,
`field_ft_125` tinytext,
`field_id_126` text,
`field_ft_126` tinytext,
`field_id_146` text,
`field_ft_146` tinytext,
`field_id_147` decimal(10,4) DEFAULT '0.0000',
`field_ft_147` tinytext,
`field_id_148` text,
`field_ft_148` tinytext,
`field_id_149` text,
`field_ft_149` tinytext,
`field_id_150` text,
`field_ft_150` tinytext,
`field_id_151` text,
`field_ft_151` tinytext,
`field_id_152` text,
`field_ft_152` tinytext,
`field_id_153` text,
`field_ft_153` tinytext,
`field_id_154` text,
`field_ft_154` tinytext,
`field_id_155` text,
`field_ft_155` tinytext,
`field_id_158` text,
`field_ft_158` tinytext,
`field_id_159` text,
`field_ft_159` tinytext,
`field_id_160` text,
`field_ft_160` tinytext,
`field_id_161` text,
`field_ft_161` tinytext,
PRIMARY KEY (`entry_id`),
KEY `channel_id` (`channel_id`),
KEY `site_id` (`site_id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;
I have an ExpressionEngine installation that was recently migrated to InnoDB from MyISAM for performance reasons. The main entry table is throwing errors when trying to update large entries:
> Row size too large (> 8126). Changing some columns to TEXT or BLOB or using ROW_FORMAT=DYNAMIC or ROW_FORMAT=COMPRESSED may help. In current row format, BLOB prefix of 768 bytes is stored inline.
The `UPDATE` query ExpressionEngine is trying to run is huge (updating around 50 columns, with a lot of large content). But this error is only ever thrown when updating one specific `TEXT` column (`field_id_13`), with data only 1500 bytes in length.
- If I comment out the setting of `field_id_13`, the query works every single time
- If I comment out `field_id_13` and add extra fields totalling more data being inserted, the query still works!
- It's only this one column that is causing the error - commenting or adding any other columns gives no problems at all
The column is being set with the following data:
`field_id_13` = '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',
That's 1540 characters of data. The data for `field_id_13` is base64-encoded data of a serialised PHP object (not my design decision unfortunately - this is done by an EE field type). When decoded, the string length is only 1154 characters.
The column `field_id_13` is set as `TEXT`, though I have also tried setting it as `VARCHAR(10000)`, `MEDIUMTEXT`, `LONGTEXT`, and `BLOB` - neither of which make any difference as the error is always thrown.
The MySQL installation is using the Antelope file system. I am aware a move can be made to Barracuda and I can use `ROW_FORMAT=compressed` but the client is very performance-conscious and I am reluctant to bring the increase in disk usage overhead for compressing/decompressing on the fly.
Any ideas what's going on?
For reference, here is my `CREATE TABLE` result:
CREATE TABLE `exp_channel_data` (
`entry_id` int(10) unsigned NOT NULL,
`site_id` int(4) unsigned NOT NULL DEFAULT '1',
`channel_id` int(4) unsigned NOT NULL,
`field_id_2` text,
`field_ft_2` tinytext,
`field_id_3` text,
`field_ft_3` tinytext,
`field_id_4` text,
`field_ft_4` tinytext,
`field_id_5` text,
`field_ft_5` tinytext,
`field_id_12` mediumtext,
`field_ft_12` tinytext,
`field_id_13` text,
`field_ft_13` tinytext,
`field_id_18` mediumtext,
`field_ft_18` tinytext,
`field_id_19` mediumtext,
`field_ft_19` tinytext,
`field_id_20` text,
`field_ft_20` tinytext,
`field_id_21` mediumtext,
`field_ft_21` tinytext,
`field_id_22` mediumtext,
`field_ft_22` tinytext,
`field_id_23` text,
`field_ft_23` tinytext,
`field_id_24` int(11) DEFAULT '0',
`field_ft_24` tinytext,
`field_id_25` mediumtext,
`field_ft_25` tinytext,
`field_id_26` text,
`field_ft_26` tinytext,
`field_id_46` float DEFAULT '0',
`field_ft_46` tinytext,
`field_id_47` mediumtext,
`field_ft_47` tinytext,
`field_id_49` text,
`field_ft_49` tinytext,
`field_id_50` text,
`field_ft_50` tinytext,
`field_id_51` text,
`field_ft_51` tinytext,
`field_id_52` text,
`field_ft_52` tinytext,
`field_id_53` mediumtext,
`field_ft_53` tinytext,
`field_id_54` text,
`field_ft_54` tinytext,
`field_id_55` text,
`field_ft_55` tinytext,
`field_id_56` text,
`field_ft_56` tinytext,
`field_id_91` text,
`field_ft_91` tinytext,
`field_id_92` text,
`field_ft_92` tinytext,
`field_id_93` text,
`field_ft_93` tinytext,
`field_id_94` text,
`field_ft_94` tinytext,
`field_id_95` text,
`field_ft_95` tinytext,
`field_id_96` text,
`field_ft_96` tinytext,
`field_id_97` text,
`field_ft_97` tinytext,
`field_id_98` text,
`field_ft_98` tinytext,
`field_id_99` text,
`field_ft_99` tinytext,
`field_id_100` text,
`field_ft_100` tinytext,
`field_id_101` text,
`field_ft_101` tinytext,
`field_id_102` text,
`field_ft_102` tinytext,
`field_id_103` text,
`field_ft_103` tinytext,
`field_id_104` text,
`field_ft_104` tinytext,
`field_id_105` text,
`field_ft_105` tinytext,
`field_id_106` text,
`field_ft_106` tinytext,
`field_id_107` mediumtext,
`field_ft_107` tinytext,
`field_id_109` text,
`field_ft_109` tinytext,
`field_id_110` text,
`field_ft_110` tinytext,
`field_id_111` text,
`field_ft_111` tinytext,
`field_id_112` text,
`field_ft_112` tinytext,
`field_id_115` text,
`field_ft_115` tinytext,
`field_id_116` text,
`field_ft_116` tinytext,
`field_id_119` text,
`field_ft_119` tinytext,
`field_id_120` text,
`field_ft_120` tinytext,
`field_id_121` text,
`field_ft_121` tinytext,
`field_id_122` text,
`field_ft_122` tinytext,
`field_id_123` text,
`field_ft_123` tinytext,
`field_id_124` text,
`field_ft_124` tinytext,
`field_id_125` text,
`field_ft_125` tinytext,
`field_id_126` text,
`field_ft_126` tinytext,
`field_id_146` text,
`field_ft_146` tinytext,
`field_id_147` decimal(10,4) DEFAULT '0.0000',
`field_ft_147` tinytext,
`field_id_148` text,
`field_ft_148` tinytext,
`field_id_149` text,
`field_ft_149` tinytext,
`field_id_150` text,
`field_ft_150` tinytext,
`field_id_151` text,
`field_ft_151` tinytext,
`field_id_152` text,
`field_ft_152` tinytext,
`field_id_153` text,
`field_ft_153` tinytext,
`field_id_154` text,
`field_ft_154` tinytext,
`field_id_155` text,
`field_ft_155` tinytext,
`field_id_158` text,
`field_ft_158` tinytext,
`field_id_159` text,
`field_ft_159` tinytext,
`field_id_160` text,
`field_ft_160` tinytext,
`field_id_161` text,
`field_ft_161` tinytext,
PRIMARY KEY (`entry_id`),
KEY `channel_id` (`channel_id`),
KEY `site_id` (`site_id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;