[Dovecot] Corrupted index & index data file
We just moved to a new box for POP/IMAP services. I rsynced /home and /var/spool/mail to the new box from the old with the archive switch. It went smoothly but I am seeing more corrupted index & index data file errors in the log than I normally do. I thought I read that Dovecot would rebuild these files after the user accessed them the second time, if corrupt. Is this correct ? Is there a better way to deal with this issue other than deleting them so they are recreated the next time the user checks their mail ?
james jedwards@nmcourts.com
On Jul 10, 2006, at 4:51 AM, james edwards wrote:
We just moved to a new box for POP/IMAP services. I rsynced /home
and /var/spool/mail to the new box from the old with the archive
switch. It went smoothly but I am seeing more corrupted index &
index data file errors in the log than I normally do.
What Dovecot version? (don't bother with 0.99) What error messages
exactly? Do you use NFS? (needs configuration changes)
I thought I read that Dovecot would rebuild these files after the
user accessed them the second time, if corrupt. Is this correct ?
Is there a better way to deal with this issue other than deleting
them so they are recreated the next time the user checks their mail ?
It usually fixes all the errors automatically by itself.
On Mon, 2006-07-10 at 08:11 -0600, james edwards wrote:
What Dovecot version? (don't bother with 0.99) What error messages exactly? Do you use NFS? (needs configuration changes)
dovecot-0.99.11-2.EL4.1
I'd suggest upgrading to 1.0rc2 and not even trying to figure out what's wrong with 0.99. It's no longer supported.
On 7/10/06, Timo Sirainen tss@iki.fi wrote:
I'd suggest upgrading to 1.0rc2 and not even trying to figure out what's wrong with 0.99. It's no longer supported.
Roger that ! I had to duild this server quick (2 days) as the old mail server was almost dead. CentOS/RHEL does not yet have a 1.0 rpm (Core 5 does). I was in a real hurry or I would of compiled dovecot by hand. If CentOS does not come out with a 1.0 rpm shortly I will upgrade by hand.
I am having problems w/locking (the old box had the same problem, so this is not due to the move):
imap(tocmtgp): Jul 10 10:35:13 Error: Error rewriting mbox file /var/spool/mail/tocmtgp: Unexpected end of file imap(tocmtgp): Jul 10 10:35:13 Error: mbox file /var/spool/mail/tocmtgp was modified while we had it read-locked, make sure all your software accessing mboxes uses the same locking methods (fcntl, flock, dotlock)
OK, if I understand things sendmail hands off the procmail to write to /var/spool/mail/<username>. Dovecot reads this file and writes all its files/dirs to /home/<username>.
So this issue is how locking is done by procmail & dovecot ?
[root@jidmail ~]# procmail -v procmail v3.22 2001/09/10 Locking strategies: dotlocking, fcntl() Default rcfile: $HOME/.procmailrc It may be writable by your primary group Your system mailbox: /var/mail/root
[root@jidmail ~]# cat /etc/dovecot.conf | grep locks # to give "mail" group write access to /var/mail to be able to create dotlocks. # with is important to avoid deadlocks if other MTAs/MUAs are using both fcntl mbox_locks = fcntl [root@jidmail ~]#
I can't find a lock file in /var/mail/ ie: ls -la | egrep *lock* or *.lock* or lock
So is procmail using fcntl or am I just not finding the lock files.
Any suggestions on fixing this ?
james
participants (2)
-
james edwards
-
Timo Sirainen