Re: 2.99.20 available

[Matti Aarnio <mea@nic.funet.fi>]

> Hello,
> 
> >From the ChangeLog in 2.99.20, 
> > 	* include/ta.h, transports/*/*.c
> > 	    Changed the way the  diagnostic() call works,
> > 	    now the SMTP transporter can send a "retryat +NNN "
> > 	    request.  It is used for rescheduling entire thread
> > 	    for a bit latter time -- now fixed at 60 seconds.
> 						 ^^^^
> 
> I found the new 2.99.20 scheduler always reschedule those undelivered smtp
> mail in 1m interval that neglecting any value in "scheduler.conf" interval
> and retries attribute.   Does it relate to your current fix ?

	Yes, because on connection failure the rescheduling step-size
	comes from the transporter, not from the scheduler, and there is
	no knowledge of scheduler config...  It could (propably should)
	be changed to some sort of flag indication, that the scheduler
	is to do the retry time pickup.

> Seems the timeout problem reported before has already fixed.  However, 
> another problem arise.  Still require much polishing before new scheduler
> can be usable ??

	It is a nuisance, but much less so than running one process
	fixed for a thread that all the time just reports error at
	connection/timeout at the initial connection..

	I have a feeling there is some error in the system, but I am
	yet to catch it "red handed", that is, happening actively,
	before I can even agree on its existence.
	("select() on bad fd", said my sysadm friend, but no written
	 report available..)


> Lai Yiu Fai                       |  Tel.:       (852) 2358-6202
> Centre of Computing Services      |  Fax.:       (852) 2358-0967
>  & Telecommunications             |  E-mail:     ccyflai@uxmail.ust.hk

	/Matti Aarnio <mea@nic.funet.fi>