Crashes, redux. I hope I have provided all the information required for a solution. Many thanks in advance for having a look.
I have 71 core files for a user, that all happened in the space of about 6 hours. It appears that mail delivered to 'Junk E-mail' is being accessed. I suspect they're all the same issue. I saw the same syslog entry a while back; did a resync & enabled process dumps. Naturally, it went away -- until this cluster of crashes.
File system is ext3. It is NFS mounted by other machines, but only the local machine should be touching the mail directories. The user does not have an interactive login - it's an e-mail only account.
This user's IMAP client is AppleMail. The delivery agent is procmail; Junk is detected by spamassassin; clamav is also present.
pop3 is configured, but I don't think anyone uses it.
Typical syslog (time matches core.20063): Dec 17 18:06:22 overkill dovecot: imap(<user>): Error: Next message unexpectedly corrupted in mbox file /home/<user>/mail/Junk E-mail at 1021601 Dec 17 18:06:22 overkill dovecot: imap(<user>): Panic: file mbox-sync.c: line 152 (mbox_sync_read_next_mail): assertion failed: (sync_ctx->input->v_offset != mail_ctx->mail.from_offset || sync_ctx->input->eof) Dec 17 18:06:22 overkill dovecot: imap(<user>): Error: Raw backtrace: /usr/local/lib/dovecot/libdovecot.so.0 [0x274ecf] -> /usr/local/lib/dovecot/libdovecot.so.0 [0x274f4d] -> /usr/local/lib/dovecot/libdovecot.so.0 [0x27446b] -> /usr/local/lib/dovecot/libdovecot-storage.so.0 [0x16633e] -> /usr/local/lib/dovecot/libdovecot-storage.so.0(mbox_sync+0x6b6) [0x168146] -> /usr/local/lib/dovecot/libdovecot-storage.so.0(mbox_storage_sync_init+0x81) [0x169391] -> /usr/local/lib/dovecot/libdovecot-storage.so.0(mailbox_sync_init+0x40) [0x186270] -> /usr/local/lib/dovecot/libdovecot-storage.so.0(mailbox_sync+0x3d) [0x18715d] -> /usr/local/lib/dovecot/libdovecot-storage.so.0(index_storage_get_status+0x74) [0x1b6c44] -> /usr/local/lib/dovecot/libdovecot-storage.so.0(mailbox_get_status+0x5f) [0x1895bf] -> dovecot/imap(imap_status_get+0x7a) [0x806396a] -> dovecot/imap(cmd_status+0x128) [0x80588a8] -> dovecot/imap(command_exec+0x33) [0x805c8e3] -> dovecot/imap [0x805b63f] -> dovecot/imap [0x805b6c3] -> dovecot/imap(client_handl Dec 17 18:06:22 overkill dovecot: imap(<user>): Fatal: master: service(imap): child 20063 killed with signal 6 (core dumped)
Tracebacks in attached archive were generated with:
cd ~<user>; for C in core.*; do gdb /usr/local/libexec/dovecot/imap $C -x gdb -q ; mv gdb.log gdb.$C.log; done
where 'gdb' is:
set pagination off set logging overwrite set logging file gdb.log set logging on bt full q Note that there are unprintable characters in the log files (binary arguments).
I have saved, but not attached the mail file & it's .lock file before doing a doveadm force-resync -u <user> 'Junk E-mail'. Let me know if you require a copy to debug this. I have also saved the core files.
Here is the dovecot config (I have replaced the domain with example.net).
dovecot -n # 2.2.15: /usr/local/etc/dovecot/dovecot.conf # OS: Linux 2.6.22.14-72.fc6 i686 Fedora Core release 6 (Zod) first_valid_gid = 4000 first_valid_uid = 4000 hostname = smtp.example.net login_greeting = Dovecot ready. Unauthorized access is prohibited. mail_access_groups = mail mail_location = mbox:~/mail:INBOX=/var/mail/%n mail_privileged_group = mail namespace inbox { inbox = yes location = mailbox Drafts { special_use = \Drafts } mailbox Junk { special_use = \Junk } mailbox Sent { special_use = \Sent } mailbox "Sent Messages" { special_use = \Sent } mailbox "Train As Forgotten" { auto = subscribe } mailbox "Train As Ham" { auto = subscribe } mailbox "Train As Spam" { auto = subscribe } mailbox Trash { auto = subscribe special_use = \Trash } prefix = separator = / } passdb { driver = pam } service imap-login { inet_listener imap {
address = imap.v4.example.net imap.v6.example.net
port = 143
} inet_listener imaps { address = imap.v4.example.net imap.v6.example.net port = 993 ssl = yes } } service pop3-login { inet_listener pop3 { address = pop.v4.example.net pop.v6.example.net port = 110 } inet_listener pop3s { address = pop.v4.example.net pop.v6.example.net port = 995 ssl = yes } } ssl_ca =
Finally, here is a list of all the core files showing their creation times. Note they're all the same size. ls -l core* | cut -d ' ' -f5- | sort 536576 Dec 17 18:01 core.20067 536576 Dec 17 18:06 core.20063 536576 Dec 17 18:11 core.20464 536576 Dec 17 18:16 core.20665 536576 Dec 17 18:21 core.20913 536576 Dec 17 18:26 core.21098 536576 Dec 17 18:31 core.21359 536576 Dec 17 18:36 core.21589 536576 Dec 17 18:41 core.21805 536576 Dec 17 18:46 core.22077 536576 Dec 17 18:51 core.22262 536576 Dec 17 18:56 core.22518 536576 Dec 17 19:01 core.22771 536576 Dec 17 19:06 core.22955 536576 Dec 17 19:11 core.23227 536576 Dec 17 19:16 core.23438 536576 Dec 17 19:21 core.23679 536576 Dec 17 19:26 core.23856 536576 Dec 17 19:31 core.24089 536576 Dec 17 19:36 core.24328 536576 Dec 17 19:41 core.24569 536576 Dec 17 19:46 core.24806 536576 Dec 17 19:51 core.24994 536576 Dec 17 19:56 core.25233 536576 Dec 17 20:01 core.25447 536576 Dec 17 20:06 core.25612 536576 Dec 17 20:11 core.25825 536576 Dec 17 20:16 core.26091 536576 Dec 17 20:21 core.26309 536576 Dec 17 20:26 core.26453 536576 Dec 17 20:31 core.26773 536576 Dec 17 20:36 core.27020 536576 Dec 17 20:41 core.27278 536576 Dec 17 20:46 core.27505 536576 Dec 17 20:51 core.27681 536576 Dec 17 20:56 core.27913 536576 Dec 17 21:01 core.28214 536576 Dec 17 21:06 core.28445 536576 Dec 17 21:11 core.28739 536576 Dec 17 21:16 core.28975 536576 Dec 17 21:21 core.29205 536576 Dec 17 21:26 core.29407 536576 Dec 17 21:31 core.29635 536576 Dec 17 21:36 core.29892 536576 Dec 17 21:41 core.30132 536576 Dec 17 21:46 core.30407 536576 Dec 17 21:51 core.30542 536576 Dec 17 21:56 core.30834 536576 Dec 17 22:01 core.31099 536576 Dec 17 22:06 core.31264 536576 Dec 17 22:11 core.31529 536576 Dec 17 22:16 core.31761 536576 Dec 17 22:21 core.31981 536576 Dec 17 22:26 core.32191 536576 Dec 17 22:31 core.32436 536576 Dec 17 22:36 core.32665 536576 Dec 17 22:41 core.449 536576 Dec 17 22:46 core.676 536576 Dec 17 22:51 core.877 536576 Dec 17 22:56 core.1134 536576 Dec 17 23:01 core.1332 536576 Dec 17 23:06 core.1516 536576 Dec 17 23:11 core.1757 536576 Dec 17 23:16 core.1935 536576 Dec 17 23:21 core.2175 536576 Dec 17 23:26 core.2380 536576 Dec 17 23:31 core.2625 536576 Dec 17 23:36 core.2871 536576 Dec 17 23:41 core.3141 536576 Dec 17 23:46 core.3460 536576 Dec 17 23:51 core.3631
-- Timothe Litt ACM Distinguished Engineer
This communication may not represent the ACM or my employer's views, if any, on the matters discussed.
On 12/18/2014 02:23 PM, Timothe Litt wrote:
Crashes, redux. I hope I have provided all the information required for a solution. Many thanks in advance for having a look.
I have 71 core files for a user, that all happened in the space of about 6 hours. It appears that mail delivered to 'Junk E-mail' is being accessed. I suspect they're all the same issue. I saw the same syslog entry a while back; did a resync & enabled process dumps. Naturally, it went away -- until this cluster of crashes.
File system is ext3. It is NFS mounted by other machines, but only the local machine should be touching the mail directories. The user does not have an interactive login - it's an e-mail only account.
This user's IMAP client is AppleMail. The delivery agent is procmail; Junk is detected by spamassassin; clamav is also present. The patches mentioned in http://markmail.org/message/xqu3yr52c6hjxqk2 might fix your issue.
You could also consider switching over to LMTP or dovecot-lda as the mail delivery method.
br, Teemu Huovila
participants (2)
-
Teemu Huovila
-
Timothe Litt