Greetings,
I recently upgraded our server from Dovecot 1.0.14 to 1.1.1. I am building using the FreeBSD ports system, so nothing is custom aside from the flags set there.
dovecot fell down a few days after the update with:
pid 1823 (imap), uid 1000: exited on signal 11
the mail log shows:
Jul 4 14:36:07 hostname dovecot: IMAP(username): Corrupted index cache file /home/username/mail/.imap/INBOX/dovecot.index.cache: Broken MIME parts for mail UID 10462
Jul 4 14:36:07 hostname dovecot: child 1823 (imap) killed with signal 11
unfortunately I do not have a core. The closest thing I could find in the Changelog is:
v1.1.rc13 2008-06-20 Timo Sirainen <tss@iki.fi> - Message parser: Fixed assert-crash if cached MIME structure was broken.
but, it does not appear to be the same issue. The upgrade file mentions issues with cache/index files, but they are related to downgrades.
The rest of the relative information:
Dovecot version: v1.1.1 Operating system: FreeBSD 6.2 p11 CPU architecture: x86 Filesystem you used: FreeBSD UFS Message that caused the crash: attached Dovecot configuration:
1.1.1: /usr/local/etc/dovecot.conf Warning: mail_extra_groups setting was often used insecurely so it is now deprecated, use mail_access_groups or mail_privileged_group instead protocols: imap imaps pop3 pop3s listen(default): 127.0.0.1:143 listen(imap): 127.0.0.1:143 listen(pop3): 127.0.0.1:110 ssl_listen(default): *:993 ssl_listen(imap): *:993 ssl_listen(pop3): *:995 ssl_ca_file: /etc/ssl/godaddy_cert/gd_intermediate.crt ssl_cert_file: /etc/ssl/godaddy_cert/hostname.cert ssl_key_file: /etc/ssl/godaddy_cert/hostname.key ssl_parameters_regenerate: 400 login_dir: /var/run/dovecot/login login_executable(default): /usr/local/libexec/dovecot/imap-login login_executable(imap): /usr/local/libexec/dovecot/imap-login login_executable(pop3): /usr/local/libexec/dovecot/pop3-login login_greeting: Hostname ready. login_process_per_connection: no login_process_size: 128 verbose_proctitle: yes first_valid_uid: 1000 first_valid_gid: 1000 mail_extra_groups: mail mail_access_groups: mail mail_location: mbox:~/mail:INBOX=/var/mail/%u mbox_read_locks: fcntl dotlock mbox_write_locks: fcntl dotlock mail_executable(default): /usr/local/libexec/dovecot/imap mail_executable(imap): /usr/local/libexec/dovecot/imap mail_executable(pop3): /usr/local/libexec/dovecot/pop3 mail_plugin_dir(default): /usr/local/lib/dovecot/imap mail_plugin_dir(imap): /usr/local/lib/dovecot/imap mail_plugin_dir(pop3): /usr/local/lib/dovecot/pop3 imap_client_workarounds(default): delay-newmail outlook-idle netscape-eoh tb-extra-mailbox-sep imap_client_workarounds(imap): delay-newmail outlook-idle netscape-eoh tb-extra-mailbox-sep imap_client_workarounds(pop3): pop3_client_workarounds(default): pop3_client_workarounds(imap): pop3_client_workarounds(pop3): outlook-no-nuls oe-ns-eoh auth default: mechanisms: plain login passdb: driver: pam userdb: driver: passwd socket: type: listen client: path: /var/spool/postfix/private/auth mode: 432 user: postfix group: postfix
I have not been able to repeat this so far.
Thanks,
John
On Mon, 2008-07-07 at 15:14 +0300, John Wood wrote:
Greetings,
I recently upgraded our server from Dovecot 1.0.14 to 1.1.1. I am building using the FreeBSD ports system, so nothing is custom aside from the flags set there.
dovecot fell down a few days after the update with:
pid 1823 (imap), uid 1000: exited on signal 11
the mail log shows:
Jul 4 14:36:07 hostname dovecot: IMAP(username): Corrupted index cache file /home/username/mail/.imap/INBOX/dovecot.index.cache: Broken MIME parts for mail UID 10462
Jul 4 14:36:07 hostname dovecot: child 1823 (imap) killed with signal 11
unfortunately I do not have a core. The closest thing I could find in the Changelog is:
Can you still reproduce this crash? http://dovecot.org/bugreport.html lists some things about how to get a core dump.
Message that caused the crash: attached
I doubt the message itself had anything to do with it. Just deleting dovecot.index.cache probably fixes the issue?
Anyway I just tried causing the "Broken MIME parts" error myself. I couldn't get Dovecot to crash (and the problem got fixed automatically).
Timo,
Can you still reproduce this crash? http://dovecot.org/bugreport.html lists some things about how to get a core dump.
I changed things to try to get a core and, of course, for nearly a 2 weeks got nothing at all. Figures. After you released 1.1.2 I went ahead and updated. A few days later I got a nice core waiting for me.
Unfortunately I made the mistake of not watching the ports build and didn't see that that it lacked -g, so the backtrace is only partially useful.
#0 0x080ba927 in message_parse_header_next () (gdb) bt #0 0x080ba927 in message_parse_header_next () #1 0x080b7f8a in imap_write_flags () #2 0x080b85c7 in imap_write_flags () #3 0x080c4dfd in i_stream_read_data () #4 0x080bd07d in message_get_body_size () #5 0x0806149a in imap_fetch_deinit () #6 0x080600c1 in imap_fetch () #7 0x0805a5ea in cmd_fetch () #8 0x0805e7ee in client_handle_unfinished_cmd () #9 0x0805e7a7 in client_handle_unfinished_cmd () #10 0x0805ea2d in client_handle_unfinished_cmd () #11 0x0805ebf1 in client_input () #12 0x080c7c30 in io_loop_handler_run () #13 0x080c71ed in io_loop_run () #14 0x08065b64 in main ()
In an attempt to stop wasting your time I have recompiled with -g and an now running that set of binaries. The next time this happens I should have useful backtrace for 1.1.2.
I doubt the message itself had anything to do with it. Just deleting dovecot.index.cache probably fixes the issue?
Anyway I just tried causing the "Broken MIME parts" error myself. I couldn't get Dovecot to crash (and the problem got fixed automatically).
I've not gone as far as doing that. All of the users on this system use mbox format, so if you think it would help I can just run a find | xargs to remove all of the dovecot.index.cache files. However, if I do this I may deny you the chance to see what keeps causing this. Would you prefer I leave it and try for another core? The crashes do not happen enough that users are screaming for my blood, yet :)
John
On Aug 8, 2008, at 1:17 AM, John Wood wrote:
Can you still reproduce this crash? http://dovecot.org/bugreport.html lists some things about how to get a core dump.
I changed things to try to get a core and, of course, for nearly a 2
weeks got nothing at all. Figures. After you released 1.1.2 I went
ahead and updated. A few days later I got a nice core waiting for me.Unfortunately I made the mistake of not watching the ports build and
didn't see that that it lacked -g, so the backtrace is only
partially useful.
It doesn't show much more than that it most likely died in FETCH
command. A lot of the function call order is impossible so the
backtrace was somehow wrong/corrupted.
In an attempt to stop wasting your time I have recompiled with -g
and an now running that set of binaries. The next time this happens
I should have useful backtrace for 1.1.2.
Thanks.
I doubt the message itself had anything to do with it. Just deleting dovecot.index.cache probably fixes the issue?
Anyway I just tried causing the "Broken MIME parts" error myself. I couldn't get Dovecot to crash (and the problem got fixed
automatically).I've not gone as far as doing that. All of the users on this system
use mbox format, so if you think it would help I can just run a find
| xargs to remove all of the dovecot.index.cache files. However, if
I do this I may deny you the chance to see what keeps causing this.
Would you prefer I leave it and try for another core? The crashes do
not happen enough that users are screaming for my blood, yet :)
Just leave them for now :)
participants (2)
-
John Wood
-
Timo Sirainen