issues

This tool was proposed in  issue 401  and first released in Maatkit 4790. It has basic functionality and tests.

Outstanding ideas include (from  issue 401 ):

1. A "kill all" mode where it just kills everything (--all) without any filtering at all except the replication thread and its own thread.
2. Moving average: "If a user or a particular query fingerprint has been abusing the server in the last window amount of time, start killing it. This tolerates one-off bad queries but does not tolerate an abusive query after a bit."

I have proposed to merge mk-kill and mk-loadavg into a single script.