Apologies if this is the wrong thread to put this question, but I feel it fills context as to what we are running.
Basically I am running a 6 node cluster which works beautifully, however I have hit a snag and I am not 100% sure how to resolve it as I have conflicting documentation and ideas.
On my initial import from dump I changed the Engine from myisam to innodb and the import worked with out fault and so does the website reading the data.
Now Fast Forward a few days. A user does an update via the CMS then pops up this error:
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.
So I start work fixing the error and the most common fix is to use baracuda table space, so off to the none production machine I go.
Fresh install Percona-Server
Add:
innodb_file_format = barracuda
innodb_file_per_table = 1
run mysql_install_db
start mysql
Confirm innodb options:
±-------------------------±----------+
| Variable_name | Value |
±-------------------------±----------+
| innodb_file_format | Barracuda |
| innodb_file_format_check | ON |
| innodb_file_format_max | Antelope |
±-------------------------±----------+
Now I import the database table I am having issues with but I see row_format to dynamic:
ERROR 1118 (42000) at line 25: Row size too large (> 8126). Changing some columns to TEXT or BLOB may help. In current row format, BLOB prefix of 0 bytes is stored inline.
Okay fine lets try row_format = compact:
That works
But wait, why does docs say use barracuda when you hit this issue , when in fact it does not, seems like its the other way around. I dont know maybe I am missing something here.
The table I am working on has over 600 fields and all are text, tinytext, float, int, varchar. Before you flame me on the varchar chnag to text, i have already tried that. Same result.