I'm not sure it's actually wrong, assigning to nirbhay_c to decide.
My basis for the bug report is the max_connections definition:
https://mariadb.com/kb/en/server-system-variables/#max_connections
max_connections Description: The maximum number of simultaneous client connections
Though a wsrep_slave_threads shows up in the process list its not actually a client connection.
I see the wsrep_slave_threads also affects the status threads_connected.
| Variable_name | Value |
| Threads_connected | 66 |
I appreciate that even async replication took two "connections" as a slave and 1 per slave as a master however it seems wsrep_slave_threads can make much more of an impact into the max_connections.
I'm in favour of preserving the definition of max_connections as client connections but alter the statusThreads_connected definition to include systems threads.
Hi Daniel,
I too think "system" threads should not be included under "connection" threads. However,
I do not feel like changing the definition of Threads_connected, would rather introduce a
new status variable representing wsrep thread count. Thanks!
> I too think "system" threads should not be included under "connection" threads.
Thank you. I hoped you see it that way too.
> However,I do not feel like changing the definition of Threads_connected, would rather introduce a
new status variable representing wsrep thread count.
works for me. 
Fix pushed to maria-5.5-galera.
thank you.
http://bazaar.launchpad.net/~maria-captains/maria/maria-5.5-galera/revision/3492