Am 2005.07.29 12:10 schrieb(en) offman@xxxxxxxxxxxxxxxx:
but the whole point is that argument goes out of the window,
because you can schedule the time of say 50 faxes without
'jitter', just by setting them to the same "TTS"
It is no problem to have 50 faxes scheduled at the same time. It
_may_ be a problem (under certain circumstances) if a job is always
requeued in the same interval.
also the fact is that , because the "TTS" is at a milisecond level on
I think time resolution is 1 second but that doesn't matter. (see
below)
the submitting of jobs, the chance of 2 remote hylafax systems :
1. having exactly the same "TTS"
2. having the same "retry time"
3. triggering the jobs at exactly the same time.
is going to be fairly remote.
Since it takes the modems some seconds to do their work the two
jobs don't need to have exactly the same TTS. I think the problem
as in my constructed example could occur when the TTS of the two
jobs are only a few seconds apart.
(My example is only a guess what the jitter may be good for.)
The other thing is i think the "jitter" is a myth!!! ( see the
last job)
I found this code in version 4.2.1:
req->tts = now + (req->retrytime != 0
? req->retrytime
: (requeueInterval>>1) + (random()%requeueInterval));
Obviously there seem to be cases when Hylafax uses requeueInterval
+/-50%, i.e. jitter.
There are other cases when Hylafax does not use intentional jitter,
e.g. when you specify a retry interval (option -I to sendfax).
Either your conditions ensure that you will not get this jitter or
maybe it occurs only in special cases.
Bodo
