Mysql hung at 41% CPU Usage

We are using 3 servers. Master Server1 syncs data to Server2 & Server3.

Server1 -- Debian -- MySQL 5.5.40
Server2 -- CentOS 6.6 -- MySQL 5.5.45
Server3 -- CentOS 6.6 -- MySQL 5.5.43

All 3 servers are :
25Core 32GB RAM CPU Intel(R) Xeon(R) CPU E5-2630 v2 @2.60GHz CPU Cores 6

Three servers works fine. In certain occasion not in any specific day time; CPU usage randomly goes up to 41% in 1 or 2 servers. Its happening once or twice a day for last 1 week. Here are my investigation.

1. Leap Second -- we disabled ntp and date is synced now.
2. Process list -- Not too many connection or suspicious query that can have such effect.
3. Ran $mysql> show slave status \G .. Found no lock
4. Once noticed "Waiting for master to send event" with time 14855 etc. Which we dropped as other times this value is negligible.
5. The one I am suspecting is "Slave has read all relay log; waiting for the slave I/O thread to update it" . Where time is under 3 or 5. This is one common message that i noticed all the time. So I tried to set innodb_write_io_thread and innodb_read_io_thread to 8.
6. After restarting mysql service everything gets stable right away. So, I cant pinpoint this to Web application.

I have looked into other posts in here and tried to optimise my.cnf file. My queries are:

1. What I could do to investigate the issue other than mentioned above?
2. Any ideas what could be the cause?
3. Any Resources that I should read which could direct me to solve this issue.

Thanks.