[Raw Msg Headers][Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: SMTP transport hangs

> > > Sometimes smtp transport process hangs and stays in the process list for
> > > a long time (days).
> >      However the trace you present below looks familiar, and
> >      I propably did fix it.  "patch1c" at my workstation.
> >       ("patch1", version "c", on top of the 2.99.48 package there)

	The naming schemes at my workstation are "challenging", perhaps
	you can find the lattest base source, and its batches :)

	Anyway, since this exchange I have rewritten the smtp alarm
	trappings, and now it works all right, sans one case which
	appeared today morning. -- Of course the process might have
	been running with old binary, though I seriously doubt that.

	I tried to figure out, why that one message had hung on
	read() of a responce within smtpwrite() of "." -- operation
	that should bail out after one hour, but had been sitting
	there for the past 4 hours..

	Weird about it was that the process did not react on SIGALRM
	at all ?  Any experience of such mal-functionality among
	you Solaris users ?  (The binary does not have blocking of
	SIGALRM, which might be sensible thing to do while doing
	malloc() processing.)

> > > Maybe the scheduler could kill letargic childs?  Something else?
> Why not make it a configurable parameter in scheduler.conf?
> I would set it to 6 - 12 hours for smtp transport...

	I have now thought of it back and forth.  It is interesting
	idea, however it is a kludge to handle a problem that sits
	somewhere else.  Although in the case I mention above, it
	might be usefull backup for non-monitored systems.

> Eugene

	/Matti Aarnio