Description:
SELECT statements against the merlin repository are increasing over time, adding around ~1000 statements per minute incrementally, per hour.
The SELECT statements have been tracked down to the 'Known Item' queries:
[root@net-sup1 mleith]# ./mysqlsla --general general.log --sort c --filter -*,+SELECT
Allowed SQL statements: SELECT
Reading general log 'general.log'.
1163114 total queries, 68 unique.
Sorting by 'c'.
__ 001 _______________________________________________________________________
Count : 495117 (42%)
Database : merlin
SELECT COUNT(*) FROM dc_known_items WHERE item_id=N AND agent_id=N
__ 002 _______________________________________________________________________
Count : 225358 (19%)
Database : merlin
SELECT item_id FROM dc_known_items WHERE agent_id=N
__ 003 _______________________________________________________________________
Count : 179048 (15%)
Database : merlin
SELECT dc_schedule.collect_id FROM dc_schedule WHERE item_id=N AND instance='S' AND server_id=N
It is not clear yet whether this is due to the replication monitoring or not.
How to repeat:
o Set up a service manager, with an agent monitoring it.
o Add a number of agents monitoring replication.
o Leave the service manager and agents alone, and observe the 'Database Activity' graph
