two further observations, which may give someone a clue:
* as the rate is increased such the the interval between requests
approaches zero,
all my experiments have show freeradius to improve its response time
such that all jumps have pretty much recovered during the later
stages of the expt
* when the freeradius server recovered from the discontinuous jumps
in response
time, and the response times climb back down they do so in steps -
BUT
these steps have widths which do not correspond with the steps of
the request rate
- for example in one exp the decaying steps in the response time are
about
8 seconds wide, whereas the request rate is ramped in in 1.0 second
steps
the server is runnign on a freebas 4.9 system, the clients on various
freebsd and linux systems.
tariq
-
List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
