[Dovecot] dovecot panic after upgrade 1.0.15 to 1.1.2
I recently upgraded dovecot from 1.0.15 to 1.1.2 and since then I am getting the following types of things in the logs:
Aug 21 17:23:13 mail1 dovecot: imap-login: Login: user=<myuser>,
method=PLAIN, rip=172.16.78.20, lip=172.16.78.22
Aug 21 17:24:01 mail1 dovecot: Panic: IMAP(myuser): file index-sync.c:
line 39 (index_mailbox_set_recent_uid): assertion failed:
(seq_range_exists(&ibox->recent_flags, uid))
Aug 21 17:24:01 mail1 dovecot: IMAP(myuser): Raw backtrace: imap
[0x80d25bb] -> imap [0x80d2def] -> imap [0x80d2719] -> imap
[0x809edb8] -> imap(index_mailbox_set_recent_seq+0x46) [0x809ee06] ->
imap [0x8083183] -> imap [0x8083621] -> imap [0x8083c3c] ->
imap(mbox_sync+0x25) [0x80840d5] -> imap [0x807b993] ->
imap(mail_index_transaction_commit+0x51) [0x80aee31] ->
imap(index_transaction_commit+0x52) [0x809fe32] ->
imap(mailbox_transaction_commit_get_uids+0x34) [0x80a2874] ->
imap(cmd_copy+0x16a) [0x805b19a] -> imap [0x806068e] -> imap
[0x8060536] -> imap [0x80606ea] -> imap [0x8060791] ->
imap(client_input+0x67) [0x8060897] -> imap(io_loop_handler_run+0x112)
[0x80daa92] -> imap(io_loop_run+0x2f) [0x80d9e4f] -> imap(main+0x61)
[0x8068d81] -> /lib/tls/libc.so.6(__libc_start_main+0xda) [0x12679a]
-> imap(__fxstat64+0x6d) [0x8059ed1]
Did I miss something in the upgrade process, or is this a bug that has yet to be fixed?
This seems to be happening to many various users for various reasons. But I can verify for Andreas M. Kirchwitz who noted the other day that one of the easiest way to produce it seems to be "saving" or "moving" between folders in (Al)Pine...
If Timo wants, I can send the dovecot.(index|index.cache|index.log) files for when this happens, but I would need to know what to send (both the folder being moved from and folder being moved to, or just one of them) and anything else needed, and the email to send them to. Sending the folder itself might be problematic, but might be possible.
If we can't get this resolved by early next week, I might have to downgrade back to the previous version...
-- Eric Rostetter The Department of Physics The University of Texas at Austin
Go Longhorns!
Quoting Eric Rostetter rostetter@mail.utexas.edu:
I recently upgraded dovecot from 1.0.15 to 1.1.2 and since then I am getting the following types of things in the logs:
The following output of "dovecot -n" has me worried...
pop3_uidl_format(default): %08Xu%08Xv pop3_uidl_format(imap): %08Xu%08Xv pop3_uidl_format(pop3): %08Xv%08Xu
Is the fact that the pop3 version is different from the imap version perhaps the cause of some/all of my problems? (The pop3 was set to be compatible with wu-imap which we used before dovecot...)
-- Eric Rostetter The Department of Physics The University of Texas at Austin
Go Longhorns!
On Thu, 2008-08-21 at 17:50 -0500, Eric Rostetter wrote:
I recently upgraded dovecot from 1.0.15 to 1.1.2 and since then I am getting the following types of things in the logs:
Aug 21 17:23:13 mail1 dovecot: imap-login: Login: user=<myuser>,
method=PLAIN, rip=172.16.78.20, lip=172.16.78.22Aug 21 17:24:01 mail1 dovecot: Panic: IMAP(myuser): file index-sync.c:
line 39 (index_mailbox_set_recent_uid): assertion failed:
(seq_range_exists(&ibox->recent_flags, uid))
Quoting Timo Sirainen tss@iki.fi:
On Thu, 2008-08-21 at 17:50 -0500, Eric Rostetter wrote:
I recently upgraded dovecot from 1.0.15 to 1.1.2 and since then I am getting the following types of things in the logs:
Aug 21 17:23:13 mail1 dovecot: imap-login: Login: user=<myuser>, method=PLAIN, rip=172.16.78.20, lip=172.16.78.22
Aug 21 17:24:01 mail1 dovecot: Panic: IMAP(myuser): file index-sync.c: line 39 (index_mailbox_set_recent_uid): assertion failed: (seq_range_exists(&ibox->recent_flags, uid))
I saw that, but the line "just comment out that line" is kind of confusing as to which line (or lines) I should comment out. Is it just the assert (and leave the return), or is it the whole conditional around the assert, or something else?
-- Eric Rostetter The Department of Physics The University of Texas at Austin
Go Longhorns!
On Thu, 2008-08-21 at 20:19 -0500, Eric Rostetter wrote:
Quoting Timo Sirainen tss@iki.fi:
On Thu, 2008-08-21 at 17:50 -0500, Eric Rostetter wrote:
I recently upgraded dovecot from 1.0.15 to 1.1.2 and since then I am getting the following types of things in the logs:
Aug 21 17:23:13 mail1 dovecot: imap-login: Login: user=<myuser>, method=PLAIN, rip=172.16.78.20, lip=172.16.78.22
Aug 21 17:24:01 mail1 dovecot: Panic: IMAP(myuser): file index-sync.c: line 39 (index_mailbox_set_recent_uid): assertion failed: (seq_range_exists(&ibox->recent_flags, uid))
I saw that, but the line "just comment out that line" is kind of confusing as to which line (or lines) I should comment out. Is it just the assert (and leave the return), or is it the whole conditional around the assert, or something else?
I meant the assert itself (or change it to an i_error("recent flag bug")). Although few clients really use recent flags, so it probably doesn't matter much what you do. :)
Anyway I'll try to look into why it's crashing. But it doesn't look like I can reproduce this with my imaptest tool, since it's been running for 2,5 hours now without this crash.
You could also see if this patch happens to help (probably not): http://hg.dovecot.org/dovecot-1.1/rev/7f5cc9e805ec
Quoting Timo Sirainen tss@iki.fi:
I saw that, but the line "just comment out that line" is kind of confusing as to which line (or lines) I should comment out. Is it just the assert (and leave the return), or is it the whole conditional around the assert, or something else?
I meant the assert itself (or change it to an i_error("recent flag bug")). Although few clients really use recent flags, so it probably doesn't matter much what you do. :)
I changed the i_assert to an i_error (so I can track when it happens, and who it happens to, still).
Anyway I'll try to look into why it's crashing. But it doesn't look like I can reproduce this with my imaptest tool, since it's been running for 2,5 hours now without this crash.
Let me know if there is anything I can do to help.
You could also see if this patch happens to help (probably not): http://hg.dovecot.org/dovecot-1.1/rev/7f5cc9e805ec
I applied it, but it doesn't appear to help judging by the number of i_error log entries from my patch...
Hopefully this will clear things up for the users (so they don't notice any problems). I'll let you know if not. :)
-- Eric Rostetter The Department of Physics The University of Texas at Austin
Go Longhorns!
participants (2)
-
Eric Rostetter
-
Timo Sirainen