LMTP and big message header

[Dmitry Alyabyev]

Hello

I see interesting problem when deliver one message to Cyrus 2.2.3
this message couldn't be delivered during long period of time with diagnostic
info "LMTP timeout after end of data". 
after looking closer I found that lmtpd began to eat memory/CPU very quick just
after receiving the whole message and finnaly terminated. 
I've never seen this behavior before and it repeats every time I do deliver this
message so I begin to believe that the message is source of the problem.

the message in mail queue is splitted to two parts (I'm using Exim) - headers
(with suffix -H) and body (with suffix -D). 
header part is slightly big - 37556 Bytes

-rw-r-  1 exim  wheel  10651 Mar 12 18:03 1B1kbU-000NXH-QY-D
-rw-r-  1 exim  wheel  37556 Mar 12 18:03 1B1kbU-000NXH-QY-H

body contains 200 lines of text and seems to be correct.
headers part contains 716 lines and the biggest part is  To: field with lots of
receipients like this:

To: xxx <[EMAIL PROTECTED]>,
 yyy <[EMAIL PROTECTED]>, yyy zzz <[EMAIL PROTECTED]>,

sure I can provide these two parts exactly as they are but not in list

any ideas why lmtpd was so confused by this message ?

-- 
Dimitry

---
Home Page: http://asg.web.cmu.edu/cyrus
Wiki/FAQ: http://cyruswiki.andrew.cmu.edu
List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html

roadmap of 2.2 branch

[Dmitry Alyabyev]

hi
any prediction about date of first release inside 2.2 branch ?
is the roadmap available ?
 
-- 
Dimitry

Re: sieve: vacation & subject

[Dmitry Alyabyev]

Ken Murchison wrote:
> Dmitry Alyabyev wrote:
> 
>> Hello
>> 
>> In case of setting up vacation the return message has 'Subject: subject' if
>> it specified like:
>> vacation "vacation text here";
>> 
>> If I write:
>> vacation :days 1 :subject "My subject here" "vacation text here";
>> The message goes with empty Subject header.
>> 
>> What did I miss ?
>> Cyrus 2.2
> 
> 
> Update your source code (either via CVS or the 2.2.1 tarball)

good point but I dont't like to patch entire cyrus
can somebody provide a patch for this ?

-- 
Dimitry

sieve: vacation & subject

[Dmitry Alyabyev]

Hello

In case of setting up vacation the return message has 'Subject: subject' if it
specified like: 
vacation "vacation text here";

If I write:
vacation :days 1 :subject "My subject here" "vacation text here";
The message goes with empty Subject header.

What did I miss ?
Cyrus 2.2

-- 
Dimitry

Re: LMTPD error on NUL characters

[Dmitry Alyabyev]

Mike Cathey wrote:

> On Mon, 2003-06-23 at 08:35, Andrzej Filip wrote:
>> Ramprasad A Padmanabhan wrote:
>> > <[EMAIL PROTECTED] >: host
>> > /var/imap/socket/lmtp[/var/imap/socket/lmtp]
>> > said: 554 5.6.0 Message contains NUL characters (in reply to end of
>> > DATA command)
>> If you use sendmail as your MTA then you can add F=1 flag to cyrusv2 mailer
>> specification.
>> 
>> MODIFY_MAILER_FLAGS(`CYRUSV2',`+1')dnl
> 
> Anyone have any idea how to work around this with postfix?

I did it by 'transport_filter = /usr/bin/tr -d \\000' on exim

-- 
Dimitry

sieve problem in 2.2

[Dmitry Alyabyev]

Howdy

I'm using 2.2 branch for a while and sometimes I see problem of making
connection to sieve
It looks like TCP session is established but no banner appears.
Normaly I solve this by commenting/uncommenting corresponding line in cyrus.conf
and doing 'killall -HUP master' twice but it still be headache.

Is it known issue ? any patch ?

FreeBSD 4.8
cyrus.conf:
sieve cmd="timsieved" listen="sieve" prefork=1

-- 
Dimitry

Re: sieve: Too many levels of symbolic links

[Dmitry Alyabyev]

Ken Murchison wrote:
> Rob Siemborski wrote:
>> 
>> On Thu, 3 Apr 2003, Dmitry Alyabyev wrote:
>> 
>> > > Yeah, you made a recursive symlink.
>> >
>> > I see :)
>> > but it was made not by me but by sieve, yeah ?
>> > so I'd like to know why it happened
>> 
>> Yes, it was made by sieve.  So we need to make sieve reject scripts named
>> "default".
> 
> Right.
> 
>>  Until then, don't call your script "default".
> 
> FYI, Websieve does this automatically when no script is present, unless
> the user overrides it.

Well, I did as John A. Tamplin suggested - set $defaultscript="default_sieve" in
websieve. And now I got:
$ ls -l sieve/domain/e/example.com/u/user/
total 4
-rw---  1 cyrus  cyrus  128 Apr  3 19:23 default_sieve.bc
-rw---  1 cyrus  cyrus  424 Apr  3 19:23 default_sieve.script

and when I send a message to the user I see:
lmtpunix[11154]: IOERROR: fstating sieve script
/usr/local/cyrus/sieve/domain/e/example.com/u/user/default.bc: No such file or
directory

???

-- 
Dimitry

Re: sieve: Too many levels of symbolic links

[Dmitry Alyabyev]

Miguel Mendez wrote:

> On Thu, 03 Apr 2003 11:16:45 +0300
> Dmitry Alyabyev <[EMAIL PROTECTED]> wrote:
> 
> Hello,
> 
>> lrwx--  1 cyrus  cyrus   10 Apr  3 10:53 default.bc -> default.bc
>^^^
>> when a message delivers there is an error in cyrus log:
>> lmtpunix[10531]: IOERROR: fstating sieve script
>> /usr/local/cyrus/sieve/domain/e/example.com/u/username/default.bc: Too
>> many levels of symbolic links
>> 
>> and sieve doesn't work according to rules of default.script
>> 
>> any ideas why it happens ?
> 
> Yeah, you made a recursive symlink.

I see :)
but it was made not by me but by sieve, yeah ?
so I'd like to know why it happened

-- 
Dimitry

sieve: Too many levels of symbolic links

[Dmitry Alyabyev]

hello

strange behavior - when I use websieve 0.61 (by Alain Turbide) with timsieved of
cyrus 2.2 I have something like this in user sieve directory:

lrwx--  1 cyrus  cyrus   10 Apr  3 10:53 default.bc -> default.bc
-rw---  1 cyrus  cyrus  424 Apr  3 10:53 default.script

when a message delivers there is an error in cyrus log:
lmtpunix[10531]: IOERROR: fstating sieve script
/usr/local/cyrus/sieve/domain/e/example.com/u/username/default.bc: Too many
levels of symbolic links

and sieve doesn't work according to rules of default.script

any ideas why it happens ?

-- 
Dimitry

Re: skiplist vs. Berkeley

[Dmitry Alyabyev]

Florian Hars wrote:

> Dmitry Alyabyev wrote:
>> Which the advantages are in skiplist comparing with BerkeleyDB ?
> 
> It doesn't show seen messages as new tree times a day.

Hmm, sounds good ...
Do you know  why it happens ?

-- 
Dimitry

skiplist vs. Berkeley

[Dmitry Alyabyev]

Hello

According to changes to the Cyrus IMAP Server since 2.1.x:
...
- The default mailbox list and seen state database formats have changed to
skiplist from Berkeley and Flat, respectively.

Which the advantages are in skiplist comparing with BerkeleyDB ?
What about size of cache that can be customised in Berkley ?

-- 
Dimitry

Re: 2.2 release ?

[Dmitry Alyabyev]

Rob Siemborski wrote:

> On Mon, 3 Mar 2003, Dmitry Alyabyev wrote:
> 
>> What is expected date of releasing cyrus-2.2 ?
> 
> Right now we're looking at sometime in the next two weeks to release
> 2.2.0-ALPHA.  It'd really make me feel a lot more comfortable to hear what
> sort of success people are having with the bytecode.  (For example,
> Christian was very helpful earlier this week tracking down an odd memory
> corruption bug).

well, based on my expirience with cyrus 2.2 there is also one bug with full hash
(don't know is it present now - haven't checked cvs for 2 weeks).

the problem is when cyrus compiled with --enable-fulldirhash and when I do
create new domain by mkimap -d my.domain.com it's created in
$configdirectory/domain/ with one-level hash and 'rehash' script doesn't solve
the problem

-- 
Dimitry

2.2 release ?

[Dmitry Alyabyev]

Hello

What is expected date of releasing cyrus-2.2 ?

-- 
Dimitry

Re: Deleting records from db

[Dmitry Alyabyev]

Sebastian Konstanty Zdrojewski wrote:

> Hello,
> 
> I have an annoying problem: my collegue deleted phisically the
> directories from the user's directory, so the system, while attempting
> to delivery a message to a mailbox returns a "I/O Error". Correct until
> here.
> 
> I need to remove the entries from the mailbox database, but I can't find
> the way. Anybody has an idea?

do reconstruct the mailbox

-- 
Dimitry

2.2 stable release (again)

[Dmitry Alyabyev]

Hello

According to previous postings - is there any changes in roadmap of 2.2 release?

-- 
Dimitry


Subject: Re: 2.2 stable release
 From: Rob Siemborski  andrew.cmu.edu>
 Date: Sat, 11 Jan 2003 15:51:14 -0500 (EST)
 Cc: marc.bigler  day.com,  andrew.cmu.edu>
 
On Sat, 11 Jan 2003, Ken Murchison wrote:

> IIRC, Rob is targeting sometime in February.  I've been real busy with
> my regular job, so I haven't had much time to tie up the remaining
> virtdomain and NNTP loose ends (mostly documentation and utilities), but
> I hope to have it done whenever Rob (and possibly Jen with the Sieve
> bytecode stuff) are ready.
>
> FWIW, I've been running the 2.2 code (virtdomains: no) since I started
> working on it (6 months ago) without any problems.

Yeah, late february sounds about right, though no guarantees, there's a
number of things left to do.

We are most likely going to go with the sieve bytecode stuff (which I'd
consider early alpha-quality in the sieve-bytecode branch at this point).

The other biggie is IPv6 support, which requires some thought about how to
do some things with master (basically, do we want master to link
libcyrus).

Also misc cleanups/stability stuff (e.g. code to convert scripts ->
bytecode, stability testing, etc)

-Rob

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Rob Siemborski * Andrew Systems Group * Cyert Hall 207 * 412-268-7456
Research Systems Programmer * /usr/contributed Gatekeeper

Old timsieved (1.0)

[Dmitry Alyabyev]

Hello

I'm using cyrus-imapd-1.6.19 + SASL 1.5.27 and happy enough. 
After some changes in my OS I've set websieve-061 to deal with timsieved. 
Before I had used an old version of websieve and that was fine.
Now I cannot login to my timsieved daemon and seems to be compatibility 
problem. I see in log:

Apr 30 15:14:16 host imapd[89519]: login: x.x.x[y.y.y.y] user plaintext
Apr 30 15:14:16 host timsieved[89520]: badlogin: x.x.x[y.y.y.y] PLAIN generic 
failure

The session looks like that:
"Cyrus timsieved v1.0.0" "SASL={PLAIN}"
AUTHENTICATE "PLAIN" {45+}
A.. there is data E=
NO "generic failure"

at the same time if I remove "+" I see:
"Cyrus timsieved v1.0.0" "SASL={PLAIN}"
AUTHENTICATE "PLAIN" {45}
"Ready for data"
A.. there is data E=
NO "generic failure"

So my question is - how deep is the difference between timsieved 2.x and 1.0
What have I fix to force websieve-061 work well ?
Unfortunately I cannot find the correct scenerio for auth with timsieved 1.0 

P.S. Please send me a copy on my email

-- 
Dimitry