simple queries taking too much time

Hi,

In our logs we see some simple queries (INSERT 1 row or DELETE 1 row by primary key) that sometime takes a few seconds to complete without knowing where is spent the time and how to reduce it :

# Thread_id: 5460576 Schema: our_schema Last_errno: 0 Killed: 0
# Query_time: 2.049189 Lock_time: 0.000099 Rows_sent: 0 Rows_examined: 1 Rows_affected: 1 Rows_read: 1
# Bytes_sent: 11
use our_schema;
SET timestamp=1423648292;
DELETE FROM our_table WHERE doc_id = 825297976;

# Thread_id: 5474371 Schema: arp_tool_ms Last_errno: 0 Killed: 0
# Query_time: 2.747146 Lock_time: 0.000063 Rows_sent: 0 Rows_examined: 0 Rows_affected: 1 Rows_read: 0
# Bytes_sent: 19
SET timestamp=1423648586;
INSERT INTO our_other_table(module_name, conf_id, pending_method, pending_args, pending_autoexec) VALUES('Focus', 0, 'getContentesnotificationfromspool', 'a:4:{s:6:\"action\";s:3:\"UPD\";s:5:\"index\";s:7:\"content\";s:4:\"type\";s:6:\"person\";s:2:\"id\";s:14:\"CNT0000005yuJC\";}', 0);

We see that only one row is accessed/inserted (use of primary key) and that the lock time is really short.

However looking at the cluster statistics doesn't give us any more information about the time lost to execute our queries. It doesn't happen all the time but we see that our applications are slowed down.

You can find the screenshot of our cluster statistics at http://pbrd.co/1MwyLoQ