[Dovecot] Maildir + NFS + multiple machines = spectacular failure

[Apps Lists]

> Of course.
> We upgraded to the last kernel available (in that moment) for RHEL4:
> 2.6.9-34.ELsmp
> The release notes mentioned that some important issues related to
> locking in client
> NFS environment were missing in the kernel included in the distribution.
>

Darn.  That's the same kernel I'm running.  Must not be my lucky day.

Steve

> Regards.
>
> Apps Lists wrote:
>
>>>Hi all.
>>>We also had problems with  client NFS locking using RHEL4, kernel
>>> 2.6.9 Some of them were solved updating the kelnel from RedHat.
>>>Hope this helps.
>>>
>>>
>>>
>>
>>Hi Luis.
>>
>>Can you tell me which kernel you upgraded to?
>>
>>Thanks.
>>
>>Steve
>>
>>
>>
>>
>>>Regards.
>>>
>>>Apps Lists wrote:
>>>
>>>
>>>
>>>>I'm running beta7 on two machines, with maildir on NFS.  I have lockd
>>>> running on all machines.  I've found that Dovecot is highly unstable
>>>> with NFS when accessing a mailbox on more than one machine at the
>>>> same time.
>>>>
>>>>Both dovecot machines have:
>>>>
>>>>mmap_disable = yes
>>>>lock_method = fcntl
>>>>
>>>>NFS is version 3, exported from a third linux machine.  All machines
>>>> are running 2.6.9 kernel.
>>>>
>>>>Any ideas what's going wrong here?
>>>>
>>>>Steve
>>>>
>>>>
>>>>*** dev4 ***
>>>>
>>>>Apr 28 09:17:40 dev4 dovecot: IMAP(jtest): Duplicate file in uidlist
>>>> file /var/mailstore/72/af/375887/Maildir/dovecot-uidlist:
>>>>1146230258.P2889Q0M407851.dev4.neonova.net:2,
>>>>
>>>>Apr 28 09:17:41 dev4 dovecot: IMAP(jtest): Corrupted transaction log
>>>> file /var/mailstore/72/af/375887/Maildir/dovecot.index.log: Extension
>>>> introduction for unknown id 0
>>>>
>>>>Apr 28 09:17:41 dev4 dovecot: IMAP(jtest): file mail-index.c: line
>>>> 883 (mail_index_sync_from_transactions): assertion failed:
>>>>(hdr.log_file_int_offset == (*map)->hdr.log_file_int_offset)
>>>>
>>>>
>>>>Apr 28 09:17:41 dev4 dovecot: child 2889 (imap) killed with signal 6
>>>>
>>>>
>>>>*** dev3 ***
>>>>
>>>>
>>>>
>>>>Apr 28 09:17:39 dev3 dovecot: IMAP(jtest): Duplicate file in uidlist
>>>> file /var/mailstore/72/af/375887/Maildir/dovecot-uidlist:
>>>>1146230254.P2888Q0M651598.dev4.neonova.net
>>>>
>>>>Apr 28 09:17:41 dev3 dovecot: IMAP(jtest): Fixed index file
>>>>/var/mailstore/72/af/375887/Maildir/dovecot.index:
>>>>first_recent_uid_lowwater 4036 -> 1772
>>>>
>>>>Apr 28 09:17:44 dev3 dovecot: IMAP(jtest):
>>>>/var/mailstore/72/af/375887/Maildir/dovecot-uidlist: next_uid was
>>>> lowered (4036 -> 1774)
>>>>
>>>>Apr 28 09:17:45 dev3 dovecot: IMAP(jtest): Duplicate file in uidlist
>>>> file /var/mailstore/72/af/375887/Maildir/dovecot-uidlist:
>>>>1146145480.P5976Q0M245663.dev3.neonova.net:2,
>>>>
>>>>Apr 28 09:17:46 dev3 dovecot: IMAP(jtest): Fixed index file
>>>>/var/mailstore/72/af/375887/Maildir/dovecot.index:
>>>>first_recent_uid_lowwater 4038 -> 1774
>>>>
>>>>Apr 28 09:17:47 dev3 dovecot: IMAP(jtest):
>>>>/var/mailstore/72/af/375887/Maildir/dovecot-uidlist: next_uid was
>>>> lowered (4038 -> 1775)
>>>>
>>>>Apr 28 09:17:48 dev3 dovecot: IMAP(jtest): Duplicate file in uidlist
>>>> file /var/mailstore/72/af/375887/Maildir/dovecot-uidlist:
>>>>1146145480.P5976Q0M245663.dev3.neonova.net:2,
>>>>
>>>>Apr 28 09:17:49 dev3 dovecot: IMAP(jtest):
>>>>/var/mailstore/72/af/375887/Maildir/dovecot-uidlist: next_uid was
>>>> lowered (4039 -> 1775)
>>>>
>>>>Apr 28 09:17:50 dev3 dovecot: IMAP(jtest): Corrupted transaction log
>>>> file /var/mailstore/72/af/375887/Maildir/dovecot.index.log: Extension
>>>> introduction for unknown id 0
>>>>
>>>>Apr 28 09:17:50 dev3 dovecot: IMAP(jtest): file mail-index.c: line
>>>> 881 (mail_index_sync_from_transactions): assertion failed:
>>>>(hdr.messages_count == (*map)->hdr.messages_count)
>>>>
>>>>Apr 28 09:17:50 dev3 dovecot: IMAP(jtest):
>>>>/var/mailstore/72/af/375887/Maildir/dovecot-uidlist: next_uid was
>>>> lowered (4039 -> 1775)
>>>>
>>>>Apr 28 09:17:51 dev3 dovecot: IMAP(jtest): Corrupted transaction log
>>>> file /var/mailstore/72/af/375887/Maildir/dovecot.index.log: Append
>>>> with UID 1775, but next_uid = 4039
>>>>
>>>>Apr 28 09:17:51 dev3 dovecot: IMAP(jtest): Corrupted transaction log
>>>> file /var/mailstore/72/af/375887/Maildir/dovecot.index.log: Append
>>>> with UID 1775, but next_uid = 4039
>>>>
>>>>Apr 28 09:17:52 dev3 dovecot: IMAP(jtest): Unexpected transaction log
>>>> desync with index /var/mailstore/72/af/375887/Maildir/dovecot.index
>>>>
>>>>Apr 28 09:17:52 dev3 dovecot: IMAP(jtest): Disconnected: Mailbox is
>>>> in inconsistent state, please relogin.
>>>>
>>>>Apr 28 09:17:52 dev3 dovecot: IMAP(jtest): Fixed index file
>>>>/var/mailstore/72/af/375887/Maildir/dovecot.index:
>>>>first_recent_uid_lowwater 4039 -> 1776
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>--
>>>+----------------------------------------------^-----------------------+
>>> | Luis Meléndez Aganzo                         ^  Email: luism at uco.es
>>> | | Servicio de Informática                      ^  Tlf:
>>> 34-(9)57-211022 | | Analista - Área de Sistemas                  ^
>>> Fax: 34-(9)57-218116 | | Universidad de Córdoba (SPAIN)
>>> ^  http://www.uco.es    |
>>> +----------------------------------------------^-----------------------+
>>>
>>>
>>
>>
>>
>>
>>
>
>
> --
> +----------------------------------------------^-----------------------+
> | Luis Meléndez Aganzo                         ^  Email: luism at uco.es  |
> | Servicio de Informática                      ^  Tlf: 34-(9)57-211022 |
> | Analista - Área de Sistemas                  ^  Fax: 34-(9)57-218116 |
> | Universidad de Córdoba (SPAIN)               ^  http://www.uco.es    |
> +----------------------------------------------^-----------------------+