Hello, list,
I'm new here and also new to Dovecot. On Friday evening (perhaps not the best moment...) I sent the following question but it seems to have been lost somehow:
Yesterday I had to reboot my server machine and I think there was an email client connected from the LAN during the reboot which I did not notice then. Now that mailbox is no longer accessible: as soon as I try to connect, it lets me wait forever while openin the folder, and cycles endlessly this message in the logs:
Jul 2 19:40:37 myhostname dovecot: imap-login: Login: user=<username>, method=PLAIN, rip=#.#.#.#, lip=#.#.#.#, TLS Jul 2 19:40:37 myhostname dovecot: IMAP(username): unlink(/var/mail/username.lock) failed: Permission denied Jul 2 19:40:48 myhostname dovecot: IMAP(username): file mbox-sync.c: line 1372 (mbox_sync_handle_eof_updates): assertion failed: (trailer_size <= 2) Jul 2 19:40:48 myhostname dovecot: IMAP(username): Raw backtrace: imap [0x80b2081] -> imap [0x80b1f99] -> imap [0x807a536] -> imap [0x807ae71] -> imap(mbox_sync+0x422) [0x807d3c2] -> imap(mbox_storage_sync_init+0x3e) [0x807d8de] -> imap(imap_sync_nonselected+0x1c) [0x806446c] -> imap(_cmd_select_full+0xc3) [0x805be83] -> imap(cmd_select+0x19) [0x805c059] -> imap [0x805d5ef] -> imap [0x805d690] -> imap(_client_input +0x74) [0x805dd64] -> imap(io_loop_handler_run+0xf0) [0x80b7e20] -> imap(io_loop_run+0x18) [0x80b7338] -> imap(main+0x478) [0x8066388] -> /lib/i686/cmov/libc.so.6(__libc_start_main+0xe5) [0xb7e36455] -> imap [0x8058561] Jul 2 19:40:48 myhostname dovecot: child 2459 (imap) killed with signal 6
(I have Debian Lenny with dovecot 1.0.15.; protocol: imaps; mailbox: mbox; lock: dotlock)
I tried to add the user dovecot to the group mail and give the group mail write permission to the mailbox. The permission errors stopped appearing but everything else is still there, including .lock file and 3 imap-login processes that resume as soon as I try to kill them.
There are also 2 unimportant mailboxes that still work. I can view the beginning of the "username" mailbox with mc inner viewer, but it hangs as I go to the end of file. Tried to copy the file so I could freely experiment with it but the copying hung up when it reached 87 % (the file is 1.1 GB). The copied part was unreadable.
Tried to
formail -ds < /var/mail/username >> /backup/usernameJul3
but with the same result.
Is there a way to repair the synchronization of the mbox? Or at least a way to recover its content?
After that I also tried to "shutdown -F -r now" but to no effect. (filesystem is ext3)
Now I'm really in trouble. Could someone, please, suggest what I could try to do to solve this?
Thanks.
On 7/5/2009 1:18 AM, Alnis Morics wrote:
(mbox_sync_handle_eof_updates)
A quick google of the above error string returns a reply from Timo asking if recent changes had fixed the problem, and it appears so, so, your best bet is to upgrade (you said you were running 1.0.15, which is pretty old)...
There have been a LOT of mbox fixes since 1.0.15, so I'd recommend going ahead and moving to 1.2, which just went stable, but at least go to 1.1.16...
--
Best regards,
Charles
On Sunday 05 July 2009 21:14:10 Charles Marcus wrote:
On 7/5/2009 1:18 AM, Alnis Morics wrote:
(mbox_sync_handle_eof_updates)
A quick google of the above error string returns a reply from Timo asking if recent changes had fixed the problem, and it appears so, so, your best bet is to upgrade (you said you were running 1.0.15, which is pretty old)...
There have been a LOT of mbox fixes since 1.0.15, so I'd recommend going ahead and moving to 1.2, which just went stable, but at least go to 1.1.16...
Thanx for advice. But shame to me -- after all the hassle, it turned out that the harddisk had a bad sector exatcly in the mbox file, and I didn't see it because I was doing all the things I wrote above, remotely. Fortunately, I managed to clone the hd with dd_rhelp. Now it's up and running again.
On 7/6/2009 6:06 AM, Alnis Morics wrote:
There have been a LOT of mbox fixes since 1.0.15, so I'd recommend going ahead and moving to 1.2, which just went stable, but at least go to 1.1.16...
Thanx for advice. But shame to me -- after all the hassle, it turned out that the harddisk had a bad sector exatcly in the mbox file, and I didn't see it because I was doing all the things I wrote above, remotely. Fortunately, I managed to clone the hd with dd_rhelp. Now it's up and running again.
I'd still upgrade if I were you... again, LOTS of mbox (and other) fixes (improvements) since 1.0.15...
--
Best regards,
Charles
On Monday 06 July 2009 14:19:00 Charles Marcus wrote:
On 7/6/2009 6:06 AM, Alnis Morics wrote:
There have been a LOT of mbox fixes since 1.0.15, so I'd recommend going ahead and moving to 1.2, which just went stable, but at least go to 1.1.16...
Thanx for advice. But shame to me -- after all the hassle, it turned out that the harddisk had a bad sector exatcly in the mbox file, and I didn't see it because I was doing all the things I wrote above, remotely. Fortunately, I managed to clone the hd with dd_rhelp. Now it's up and running again.
I'd still upgrade if I were you... again, LOTS of mbox (and other) fixes (improvements) since 1.0.15...
Sure. But, as concerns mbox, I already thought of migrating to Maildir. In this situation, I would just had lost one or a couple of mail messages instead of rendering the whole mailbox unreadable.
Sorry if the question has been answered before, but I haven't been able to find any relevant topic on your mailing list archive.
We are using Dovecot 1.1.6 with legacy mbox storage on a mail server of ours; the trouble is, users connected via DSL with both a computer and a iPhone on the same wireless network experience broken connections - this is very likely a file locking issue, and as such might happen with other broken clients reliying on persistent connections.
My question: is there a way to tell Dovecot the lock is owned _only_ by the latest connection opened by the same user, therefore closing all the others?
Thanks a lot for both your help and for making such a great POP3/IMAP client available
Federico Bianchi
Dipartimento di Storia delle Arti
Universita` di Pisa
piazza S.Matteo in Soarta, 2 - I56127 Pisa (Italy)
fax.+39-050-580128; e-mail: <f.bianchi@arte.unipi.it>
===================================================
!DISCLAIMER!: my e-mail reflects _my_own_ opinions!
===================================================
Sorry if the question has been answered before, but I haven't been able to find any relevant topic on your mailing list archive.
We are using Dovecot 1.1.6 with legacy mbox storage on a mail server of ours; the trouble is, users connected via DSL with both a computer and a iPhone on the same wireless network experience broken connections - this is very likely a file locking issue, and as such might happen with other broken clients relying on persistent connections.
My question: is there a way to tell Dovecot the lock is owned _only_ by the latest connection opened by the same user, therefore closing all the others?
Thanks a lot for both your help and for making such a great POP3/IMAP client available
Federico Bianchi
Dipartimento di Storia delle Arti
Universita` di Pisa
piazza S.Matteo in Soarta, 2 - I56127 Pisa (Italy)
fax.+39-050-580128; e-mail: <f.bianchi@arte.unipi.it>
===================================================
!DISCLAIMER!: my e-mail reflects _my_own_ opinions!
===================================================
Le 6 juil. 09 à 13:29, Federico Bianchi a écrit :
Sorry if the question has been answered before, but I haven't been
able to find any relevant topic on your mailing list archive.We are using Dovecot 1.1.6 with legacy mbox storage on a mail server
of ours; the trouble is, users connected via DSL with both a
computer and a iPhone on the same wireless network experience broken
connections - this is very likely a file locking issue, and as such
might happen with other broken clients relying on persistent
connections.My question: is there a way to tell Dovecot the lock is owned _only_
by the latest connection opened by the same user, therefore closing
all the others?
Hello Frederico,
Which protocol do the client use? Both POP? Both IMAP? Or POP and IMAP?
Don't you see anything in the logs related to those "broken
connections"? And how are those failures shown by the clients?
On what kind of storage are the mailboxes stored?
And could you show us the output of dovecot -n?
Axel
participants (5)
-
Alnis Morics
-
Axel Luttgens
-
Charles Marcus
-
Federico Bianchi
-
Federico Bianchi