Tracking the wrong main memory data.

[Ben Kietzman]

Chris Weaver from AT&T reported the following (AT&T specifics have been hidden):

Received this alert:

system@******(11:00:18 AM): (chatbot,Central Monitor) ******: using 90% main memory which is more than the maximum 90%

This was status:
******@******:~$ free -m
             total used free shared buffers cached
Mem: 32175 31986 189 65 373 28269
-/+ buffers/cache: 3343 28832
Swap: 8187 0 8187

The top row 'used' (31986) value will likely nearly match the top row mem value (32175) as spare memory is used for cache (28269)

I think you should be generating alerts on the buffers/cache row used value (3343) and free (28832).