[gomd-devel] <DAEMON>: starting working on Black Hole Detection (BHD)

[Gian Paolo Ghilardi]

Hi all.

I've started working on the Black Hole Detection (BHD) feature for the GOMD
daemon.

For people ignoring what BHD is,... some words! :)
BHD is a facility that gets the status of a node and returns if the node is
working fine or malfunctioning, overloaded,... Instead of using "classic
thresholds", it uses a small Fuzzy-Logic engine to make decisions. (a
"classic threshold"==an alarm value: each time this value is trespassed, an
alarm will be raised; Fuzzy Logic is fault-tolerant as it works by "ranges"
of values and each range can overlap the others as values).

Actually I've coded a first rudimental example classifying the status of a
CPU by its temperature (i.e.: not running, cold, fresh, normal_cond, hot,
too_hot, dying).

The idea is to provide this mechanism for a lot of "entities"  (cpu
temperature, memory status, ...), combine the results of each "entity" and
decide if we need to raise an alarm or avoid doing that. I need help for
choosing these "entities": what could be important to check?  :)))

Thanks.

CU.

<rejected>