On Tue, 2006-03-21 at 13:54 +0000, Jim Wight wrote:
> The most recent run has reported as follows:
>
> *: Host x.x.x.x (hailed us) last at Sun Mar 19 23:39:41 2006
> i.e. not seen for 37.63 hours
> (Expected <delta_t> = 5184.63 secs (= 1.44 hours) (Expires 90 days)
>
> However, there is another entry in the database that I would have
> expected to have been flagged, given the following:
>
> # cfshow -s | egrep -v '!0\.|!1\.'
> x.x.x.x (hailed us) at [Sun Mar 19 23:39:41 2006] i.e. not seen for !37.76!
> hours; <delta_t> = {1.44} hours
> y.y.y.y (hailed us) at [Wed Mar 15 17:28:52 2006] i.e. not seen for !139.94!
> hours; <delta_t> = {0.49} hours
>
> Both of those should have been reported, shouldn't they?
I reckon the code that is causing the effect is the section at line 584
in ip.c:
if (average < 1800) /* anomalous couplings do not count*/
{
criterion = false;
}
I don't pretend to understand the logic, but when running cfagent with
-v, the entry that I consider is being missed has an 'Expected
<delta_t>' of less than 1800 - as do all of those which have been seen
within the specified period. The entry that is being reported correctly
has an 'Expected <delta_t>' greater than 1800.
Shouldn't 'Expected <delta_t>' only come into play if FriendStatus's
argument is zero? Normally, I would expect the actual time since last
seen to be used.
Jim
_______________________________________________
Help-cfengine mailing list
Help-cfengine@cfengine.org
http://cfengine.org/mailman/listinfo/help-cfengine
