We have a few FreeBSD hosts running Exim with something like
`smtp_load_reserve = 30` configured and occasionally see dropped
connections because of high load:
2022-02-11 10:02:07 temporarily refused connection from [1.2.3.4]: not
in reserve list and load average = 30.74
But according to our monitoring system (which tracks 1min and 15min
avg), the affected hosts never exceed an average load spike of 20.
What time span does Exim use for the average load calculation?
Could it be possible, that it uses a shorter time window and there are
short load spikes exceeding the limit of 30 or is there a possible bug?
