Just a remark: this issue (Panic: file buffer.c: line 307
(buffer_set_used_size): assertion failed: (used_size <= buf->alloc)) was
reported by me on 2014-01-11 on 2.2.10, it still occurs on 2.2.13.
Indeed, we also switched from sparc server to x86. We also removed all
the dovecot indexes. We use ma
Please explain the following dovecot (2.2.13) behaviour: I run dovecot
in ldap auth debug mode and log file excerpt follows:
2014-05-21T08:22:01+02:00 dovecot/ip dovecot: auth: Debug:
ldap(user@domain,pcip,): bind search:
base=o=tree,dc=do,dc=ma,dc=in
filter=(&(objectClass=MailAccount)(accountActi
Timo Sirainen - 2014-05-09 15:27:
>
> I thought this was fixed already, but I guess not. It would be helpful if you
> could either
>
> a) Enable debug symbols so the backtrace would show exactly where it crashes
Here you are:
Core was generated by `dovecot/imap postlogin'.
Program terminated wit
Charles Marcus - 2014-05-14 19:26:
> On 5/14/2014 12:29 PM, Maciej Uhlig wrote:
>> Reindl Harald - 2014-05-14 17:04:
>>> the mail cient on the receiver shows the FROM-HEADER and not
>>> the Return-Path
>> I'm actually looking at your mail using the newest
Reindl Harald - 2014-05-14 17:04:
> the mail cient on the receiver shows the FROM-HEADER and not
> the Return-Path
I'm actually looking at your mail using the newest Thunderbird and guess
what I see? I see your mistake:
Return-Path:
Thanks.
MU
smime.p7s
Description: Kryptograficzna sygn
Stephan Bosch - 2014-05-13 21:02:
> Keep in mind that the From: header is set to something useful
> (http://tools.ietf.org/html/rfc5230#section-5.4). That is what the
> recipient will normally see.
Right. I was looking from the mail admin point of view (there was only
"from=<>" in receiving MTA l
Reindl Harald - 2014-05-13 21:00:
> guess what happens if two persons have a vacation
> resonder at the same time and one of them sends
> a mail - the faster server with more storage will
> win that battle
Not necessarily. 'sieve_vacation_default_period' would be taken into
account.
> may i ask to
Charles Marcus - 2014-05-13 20:20:
>
> No, no, no, null sender is *required* for these kinds of messages.
>
Why? It's _my_ message after all. IMHO I'm sending "I'm on vacation",
not MTA.
Thanks.
MU
smime.p7s
Description: Kryptograficzna sygnatura S/MIME
May I ask to add Return-Path: some meaningful header line to sieve sent
mail headers in vacation message? Now the header line isn't generated at
all and the effect is as follows:
on receiving MTA:
2014-05-09T15:04:32+02:00 host/ip postfix/qmgr[2408]: 41F2F6024E:
from=<>, size=900, nrcpt=1 (queue
Reindl Harald - 2014-05-13 14:01:
> yes, it sends a *NEW* message via the MTA
OK, I need just this *NEW* mail queue id.
> but that don't change the fact the dovecot don't know
> the queue id of the *original* message received by
> the MTA
You're right - but this is something I don't need: this i
Reindl Harald - 2014-05-13 13:30:
> * from where do you think dovecot-lmtpd shoukd get the MTA's queue id?
AFAICS dovecot sieve plugin is able to send mail (vacation response,
forward mail) via SMTP MTA.
Thanks.
MU
smime.p7s
Description: Kryptograficzna sygnatura S/MIME
May I ask to add receiving smtp server mail queue id to sieve 'sent
vacation response' log line?
Now:
2014-05-09T15:04:32+02:00 name/ip dovecot: lmtp(8164, user1@domain):
YvJFLVbSbFPkHwAASpgOhg: sieve:
msgid=<000301cf6b87$36be4570$a43ad050$@us.edu.pl> size=46850: sent
vacation response to
Asked
Sorry for replying to myself but now I'm able to show gdb backtrace with
debugging symbols for dovecot 2.2.13 crash. See below.
Thanks.
MU
Maciej Uhlig - 2014-05-09 10:59:
> Timo Sirainen - 2014-05-08 17:37:
>> BTW. I've read most of the mails again in Dovecot mailing list
Teemu Huovila - 2014-05-09 12:51:
> Could you post the output of "bt full", though judging by your post in
> january, you do not have the symbols to go with it.
Right.
>
> Please also post your doveconf -n
# 2.2.13.rc1: /dir1/dir3/dovecot/dovecot.conf
# OS: Linux 2.6.32-431.1.2.0.1.el6.x86_64 x
Timo Sirainen - 2014-05-08 17:37:
> There have been a ton of smaller and some larger changes since v2.2.12, so I
> thought I'd first make a RC release and the final v2.2.13 tomorrow. Please
> try it out!
Actually there were seven identical segfaults for the same user in three
minutes:
Program
Timo Sirainen - 2014-05-08 17:37:
> BTW. I've read most of the mails again in Dovecot mailing list, but there are
> still a few reported bugs I decided to leave until later. Anyway feel free to
> re-send any bug reports for things that haven't been fixed in v2.2.13 yet.
First report on 2014-01-1
Kim Johansen - 2014-03-02 20:09:
> - but I have had success with using a bash script instead of a PHP
script, with a smaller foodprint as Jeroen suggested.
> So I think that is the solution is to rewrite the script to bash script.
Some time ago I tried to get postlogin PHP script working without
This is log from dovecot 2.1.17 (I had to downgrade from 2.2.10 because lot of
problems) on CentOS 6.5:
2014-01-14T18:36:03+01:00 server/a.b.c.d dovecot: lmtp(5927): Fatal: master:
service(lmtp): child 5927 killed with signal 11 (core dumped)
# uname -a
Linux server 2.6.32-431.1.2.0.1.el6.x86_6
This is log from dovecot 2.2.10:
2014-01-13T15:34:39+01:00 server/a.b.c.d dovecot: imap(user@domain): Fatal:
master: service(imap): child 6290 killed with signal 11 (core dumped)
(gdb) bt full
#0 0x7fe78683aba5 in mail_cache_header_fields_read ()
from /usr/local/dovecot-2.2.10/lib/doveco
This is log from dovecot 2.2.10:
2014-01-11T09:55:14+01:00 server/a.b.c.d dovecot: imap(user@domain):
Error: Corrupted transaction log file
/pracmail/domain/user/mail/.sent-mail/dovecot.index.log seq 16777216:
log file shrank (868 < 6144) (sync_offset=6144)
2014-01-11T09:55:14+01:00 server/a.b.c.d
Stephan Bosch:
>
> So, this should work for you:
>
> deliver_log_format=msgid=%m size=%p: %$
>
Oh yes it did the trick :-)
Thank you and best regards,
MU
smime.p7s
Description: Kryptograficzna sygnatura S/MIME
This is a dovecot sieve log line:
2012-09-26T13:47:52+02:00 host/192.168.1.1 dovecot: [ID 583609
mail.info] lmtp(7565, user@domain): vTtJDTrrYlCNHQAAs1WjhA: sieve:
msgid=: stored mail into mailbox 'INBOX'
Would it be possible to have mail size logged here?
MU
smime.p7s
Description: Kryptograf
W dniu 2011-06-06 00:56, Stephan Bosch pisze:
If you have a sample message, you can use the sieve-test tool for
that. You can check the individual steps the interpreter takes while
processing the message. Refer to the man page for more info.
Thanks. It could be really helpful in case of sc
W dniu 2011-06-05 11:37, Stephan Bosch pisze:
What is your application?
Actually it's something like this:
>>>
if anyof (
# example #1
allof (
...
anyof (
...
)
),
# example #2
allof (
...
anyof (
...
)
),
# example #3
allof (
...
an
Could it be possible to trace Sieve script using method other than
"enotify", i.e. to write a message to syslog using some action?
MU
15-lda.conf:
submission_host = 192.168.m.n:587
2011-03-21T14:43:49+01:00 prac/p-prac dovecot: [ID 583609 mail.debug]
lmtp(12354, u...@us.edu.pl): Debug: mcQ1NitVh01CMAAAuqlTtQ: sieve:
executing script from /mails/sieve/u...@us.edu.pl.svbin
2011-03-21T14:43:49+01:00 prac/p-prac dovecot: [ID 583
W dniu 2010-10-26 16:03, Timo Sirainen pisze:
On Tue, 2010-10-26 at 16:00 +0200, Maciej Uhlig wrote:
I'd like to write post-login script using PHP. I noticed every shell
post-login script ends with: exec "$@", which effectively executes
dovecot's script-login.
I'd like to write post-login script using PHP. I noticed every shell
post-login script ends with: exec "$@", which effectively executes
dovecot's script-login.
However, while I try to do the same (?) in PHP: $prog = $argv[1]; exec
("$prog"); it doesn't work. The client timeouts, dovecot (2.0.6
On 2010-09-23 17:08, Timo Sirainen wrote:
On Thu, 2010-09-23 at 15:33 +0100, Timo Sirainen wrote:
On Thu, 2010-09-23 at 15:00 +0200, Maciej Uhlig wrote:
2010-09-23T09:25:23+02:00 st1/192.168.10.201 dovecot: [ID 583609
mail.error] master: Error: service(lmtp): child 15524 killed with
user homedir was defined, the error disappeared.
This is under Solaris 10 with dovecot 2.0.3 and pigeonhole 0.2.0.
Thank you,
Maciej Uhlig
Timo Sirainen wrote:
Can you try if the attached patch fixes this
We did try. Unfortunately the problem is still there. Same error.
Best regards,
MU
let me provide some more findings:
first, the script which uses >>require "body"<< crashes too for some
mails while running as a script for individual user.
second, we found the test mail example which causes this crash every time.
please let me know if you're interested in the mail source, I
Hi,
there is a line in TODO: "- sieve-cmu.c crash: i_assert(buf->used - 1 ==
part->body_size.physical_size); ". Yes, we've really seen this crash.
Therefore we upgraded to dovecot-1.2.3 with dovecot-1.2-sieve-0.1.11 in
order to use sieve plugin (i.e. not cmu-sieve). We use Solaris 10 on Sun
s
Charles Marcus:
no?
Should work with both IMAP _and_ POP3.
Best regards,
MU
Timo Sirainen:
What exactly are they? I guess a global file that is served as a message
for all users?
Right. With POP3 it is served once - after first user login after the
message was placed in bulletin database (just a plaintext file in a
directory, for instance).
That's more difficult to
As a long term user of qpopper with bulletins, I'd like to know if the
bulletin feature is planned for dovecot? I really miss bulletins :-),
this is very convenient way of providing information to users.
Best regards,
MU
Timo Sirainen:
But the point is
that plaintext authentication (PLAIN or LOGIN auth mechanism) can verify
the password against ANY schema.
I see. Many thanks for this explanation :-)
Best regards,
MU
Timo Sirainen:
If the password is the same in both cases, you can simply use a single
CRAM-MD5 scheme. Dovecot can do plaintext authentication against all
schemes just fine.
Actually I happen not to understand the above :-( I thought PLAIN is a
plaintext schema while CRAM-MD5 is non-plaintex
Timo Sirainen wrote:
I've implemented it (or actually something that's a bit more generic)
this far: http://dovecot.org/patches/1.2/auth-multi-password.diff
But IIRC it's still missing things and I decided it's not worth the
effort right now.
I see. FYI I'd like to allow an user to authentica
Timo wrote : "In future it's possible that Dovecot could support
multiple passwords in different schemes for a single user.".
Is it planned in the nearest future? :-)
Best regards,
MU
40 matches
Mail list logo
