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

Re: smtpd or scheduler problem



On Wed, Nov 15, 2000 at 10:54:44PM +0100, Tomasz Bojakowski wrote:
> Hi.
> 
> First of fall thanks for the tip with ndbm :)
> And now.. I've upgraded to recent cvs shot...
> Unfortunetly a little problem occured [maybe sth. changed in config,
> didn't had time to check them yet]. Well the problem is, when for
> some reason mail couldn't be deliver to reciepient (ie. host is not
> responding) the mail lands in mail queue... everythings fine, but
> after upgrade sth. has broken.
> Mail lands there, but: a) without reason (ie. no route to host). b) tries=0
> _always_ (the retry in counts down, reaches 0, and again...).

   Yes, it used to be such that the scheduler counted the tries when
   a message was submitted into the try, now diagnostic replies are
   driving the "tries" counter -- TA programs tell when something
   really is tried.

   OTOH..  Having the counter growing fast is also a sort of sign that
   something is very much amiss, e.g. when the scheduler clocks the
   "tries" counter when the message is sent into processing.

> Oh, in logs from scheduler sth. like that:
> 
> Resyncing file "N/374569-30123" (ino=374569 pid=8913 of=1 ho='poczta.wp.pl')
> reqcnt=2
> reclaim(6,-1) pid=-8913, reaped=1, chan=smtp, host=poczta.wp.pl SIGNAL=11

  The very current cvs version *may* have problems, although I don't see
  any at my workstation, the partial rewrite of  mimeheaders.c  header
  handling may have SIGSEGV material inside.

  Do you see any dropped cores ?  Could you make sure you have compiled
  your binaries with '-g' option, so that gdb backtrace would give sensible
  callback list, including files/linenumbers of functions.

> Weird.
> os: openbsd.2.7 i386
> Oh.. and what more interesting, it happens _only_ to emails wich are send via
> smtp.. not invoked from local account (via sendmail em.)
> 
>         Tomasz Bojakowski

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