[Dovecot] corruption and errors in dovecot-stable

[Nicholas Riley]

On Mar 12, 2005, at 2:53 PM, Timo Sirainen wrote:

> On Fri, 2005-02-25 at 13:48 -0600, Nicholas Riley wrote:
>> dovecot: Feb 21 11:40:06 Error: IMAP(njr): mbox sync: UID inserted in
>> the middle of mailbox /usr/home/njriley/mail/lists/trac (5308 > 1151)
>
> dovecot-stable may handle mboxes a differently than 0.99.x, so I guess
> these could happen (although they shouldn't.. I'd like to know if this
> can be reproduced somehow?). In any case Dovecot should fix these and
> you shouldn't see these more than once per each mailbox. If they appear
> constantly for same mailboxes, there is a problem..

Since I wrote this, I have upgraded to 1.0-test63; see my other mail 
for why I can't compile -test65 :-).  I am only running this for my own 
mailbox, not for the other users on the system, who are back on 
0.99.13.

dovecot: Mar 15 22:00:12 Error: IMAP(njr): mbox sync: UID inserted in 
the middle of mailbox /usr/home/njriley/mail/lists/trac (14238 > 1151)
[...]
dovecot: Mar 16 13:02:45 Error: IMAP(njr): mbox sync: UID inserted in 
the middle
  of mailbox /usr/home/njriley/mail/lists/trac (15539 > 1151)
[...]
dovecot: Mar 16 15:47:37 Error: IMAP(njr): mbox sync: UID inserted in 
the middle of mailbox /usr/home/njriley/mail/lists/trac (16855 > 1151)

so it looks like for at least one mailbox, it does recur.

The only other message I've been getting with 1.0-test63 has been:

dovecot: Mar 15 22:00:18 Error: IMAP(njr): Fixed index file 
/usr/home/njriley/mail/.imap/Deleted Messages/dovecot.index: 
recent_messages_count 1 -> 0
[...]
dovecot: Mar 16 17:19:18 Error: IMAP(njr): Fixed index file 
/usr/home/njriley/mail/.imap/Deleted Messages/dovecot.index: 
recent_messages_count 2 -> 0

>> dovecot: Feb 25 12:30:40 Error: IMAP(sachs): Corrupted transaction log
>> file /home/sachs/Mail/.imap/INBOX/dovecot.index.log: Extension record
>> update update without intro prefix
>
> Strange. Is dovecot running over NFS or something?

No, mailboxes are on a local disk (FreeBSD 4.10 UFS).

> Can you reproduce any of these problems easily or do they occur only
> randomly?

I am loath to expose the users to these failures, but I will try again 
to test for a short period with 1.0-test65 or later once I get it to 
compile.

-- 
Nicholas Riley <njriley at uiuc.edu>     | 4111 Siebel Center
Department of Computer Science        | 201 N. Goodwin Ave.
and Medical Scholars Program          | Urbana, IL 61801-2302
Univ. of Illinois at Urbana-Champaign | +1 217 244 2274