[Dovecot] imap failed asssert in imap-bodystructure.c

[Matt Ronge]

One thing I forgot to mention: I'm running 1.0rc15

On Dec 21, 2006, at 11:21 PM, Matt Ronge wrote:

> I'm running Dovecot on a Debian system which recently has been  
> failing on an assert(). I have a spam mailbox that if I try and  
> open (from any client), the imap process dies and this is written  
> to the error logs:
>
> Dec 21 18:57:11 theronge imap(mronge): file imap-bodystructure.c:  
> line 278 (part_parse_headers): assertion failed: (part- 
> >physical_pos >= input->v_offset - start_offset)
> Dec 21 18:57:11 theronge dovecot: child 28911 (imap) killed with  
> signal 6
>
> I had this same issue with my Inbox a few weeks ago, so I moved all  
> the e-mail out and moved each piece back in one by one, in order to  
> determine if a piece of mail was causing the issue. As soon as all  
> my mail was moved back in, the issue went away and I was able to  
> load my Inbox as usual. I still have the same issue with my spam  
> mailbox, is there anything particular I can do to help debug this  
> issue? In the past I've rm'ed all the mail in my spam mailbox and  
> that fixes it.
>
> One thing that may be related is that I only noticed this issue  
> after I had setup Spambayes, which does it's work via procmail. The  
> mailboxes that I've had the issue with, all had headers that  
> contain a Spambayes header, usually like so:
>
> X-Spam-Score: 26
> X-Spam-Score: 0.6 (/)
> Sender: mailman-bounces at acm.uiuc.edu
> Errors-To: mailman-bounces at acm.uiuc.edu
> X-Spam-Score: 0.1 (/)
> X-Authenticated-Sender:
> X-Spam-Details: rule=cautious-plus_notspam policy=cautious-plus  
> score=26
>         mlx=26 adultscore=0 adjust=0 reason=mlx  
> engine=3.1.0-0611300000
>         definitions=main-0612060023
> X-Spam-OrigSender: mailman-bounces at acm.uiuc.edu
> X-Spam-Bar: ***
> X-Spambayes-Classification: spam; 1.00
> X-Spambayes-Trained: spam
>
> I'm not sure if it's related, but it's a shock in the dark on my  
> part. Any suggested steps that might help isolate this bug?
> --
> Matt Ronge
>