6 node galera cluster conflict cert failure

Question

We have a 6 node galera cluster and following table:

mysql> show create table sessions;
| Table    | Create Table
+----------+--------------
| sessions | CREATE TABLE `sessions` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`session_id` varchar(255) NOT NULL,
`data` text,
`created_at` datetime DEFAULT NULL,
`updated_at` datetime DEFAULT NULL,
PRIMARY KEY (`id`),
KEY `index_sessions_on_session_id` (`session_id`),
KEY `index_sessions_on_updated_at` (`updated_at`)
) ENGINE=InnoDB AUTO_INCREMENT=260176483 DEFAULT CHARSET=utf8 |


mysql> desc sessions;
+------------+--------------+------+-----+---------+----------------+
| Field      | Type         | Null | Key | Default | Extra          |
+------------+--------------+------+-----+---------+----------------+
| id         | int(11)      | NO   | PRI | NULL    | auto_increment |
| session_id | varchar(255) | NO   | MUL | NULL    |                |
| data       | text         | YES  |     | NULL    |                |
| created_at | datetime     | YES  |     | NULL    |                |
| updated_at | datetime     | YES  | MUL | NULL    |                |
+------------+--------------+------+-----+---------+----------------+
5 rows in set (0.00 sec)

We see many wsrep_local_cert_failures on the nodes like this:

SHOW status like '%wsrep%';
| wsrep_local_cert_failures | 165419

galera debugging displays many conflicts:

THD: 251130, mode: local, state: executing, conflict: cert failure, seqno: 92044718

Binary logging is disabled. I can identify the queries with the thread id in the general log file:

251130 Query SHOW FIELDS FROM sessions
251130 Query SELECT sessions.* FROM sessions WHERE sessions.session_id =   
'3d1d7f8638dbfd12ee58fa78d4f0998c' LIMIT 1
251130 Query BEGIN
251130 Query INSERT INTO sessions (session_id, data, created_at,  
updated_at) VALUES ('3d1d7f8638dbfd12ee58fa78d4f0998c',  
'BAh7BkkiDnJldHVybl90bwY6BkVGIgYv\n', '2016-01-04 10:48:52', '2016-01-04 
10:48:52')
251130 Query COMMIT

The Application generates the session id. Any idea?
Whats wrong with session id's and how to resolve the conflicts.

Thanks

Answer 1

Swapping on the hosts was the reason. We have no errors after setting the correct innodb_buffer_pool_size and restarting.

Thanks for helping.