Invalid postfix queue files (UNCLASSIFIED)

[Glenn Steen]

On 12/06/07, Glenn Steen <glenn.steen at gmail.com> wrote:
> On 12/06/07, Gerard <gerard at seibercom.net> wrote:
> > On June 12, 2007 at 12:13PM Glenn Steen wrote:
> >
> > {snip]
> >
> > > I'll be looking hard at this, time permitting, over the next few days.
> > > Ie also gotten a good report (very detailed) from Nerijus Baliunas
> > > along the same lines, but... I can't really see how any MailScanner
> > > code could be responsible for damaging anything in hold (might be me
> > > needing new glasses:-) simply due to the fact that we only _read_ from
> > > hold (well, eventually we unlink that message, after successful
> > > requeue, of course).
> > >
> > > But I will make an effort to ascertain whether this really is due to
> > > the code added to 4.60.8 or not. Who knows, perhaps I'm doing
> > > something stupid when droping the message from the batch (rightly)...
> > > But I rather doubt it:-)
> > >
> > > You can help further by giving more complete log snippets/examples
> > > (from connect until it pops up as invalid).
> >
> > I thought I saw some chatter regarding this on the Postfix forum a
> > short time ago. In any case, I think it can be safely stated that Wietse
> > will not be rushing to your door offering assistance.
> >
> That is todays understatement, methinks:-)
>
> And as it turns out, he likely shouldn't either.
> Nerijus helped me pinpoint that this only happens with messages
> lacking a body, so ... I need to fix that:). Unless Jules sees this
> and beats me to it, I should have something by sometime tomorrow...
> Stay tuned....:-)

As promised, here is a patch to cure MailScanner deeming "bodyless"
messages to be invalid. It's really not that massive (just an if
statement), but looks worse because of indentation (which is probably
wrong anyway... I've been a tad lazy. Again:-).
When you've applied this, the supposedly invalid messages will clear
out quickly. Remember to restart after apply.
This should be in the next beta/stable.
Hopefully (fingers crossed and all that:-) this is the final bug in
this code:-P.

> I've also been thinking of "reintroducing" pristine p records (0
> variant) wherever I munge them, simply to preserve them for any
> locally resubmitted queue files (on systems using that type of
> milter)... Just to be sure...:-). Perhaps not the most needed thing,
> so ... we'll see.

The patch doesw _NOT_ do the reintro thing. I've no time to test that. Sorry.
I imagine we'll see eventually if it is needed:).

Cheers
-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Postfix.pm.prec_fix_nobody.patch
Type: application/octet-stream
Size: 6803 bytes
Desc: not available
Url : http://lists.mailscanner.info/pipermail/mailscanner/attachments/20070613/7b5a7a5f/Postfix.pm.prec_fix_nobody.obj