[Dovecot] Wait for interface to become available instead of dying?

2013-06-10 Thread Sebastian Arcus
At the moment, if one of the interfaces specified with listen= in 
dovecot.conf is not up when Dovecot is started, then Dovecot just 
refuses to start. Is there an option to make Dovecot start anyway, and 
just use the interface when it becomes available? It is inconvenient to 
have Dovecot refuse to start during boot because some interface is 
temporarily not available.


Then again, maybe there is some strong security reasoning behind the way 
Dovecot behaves at the moment?


Re: [Dovecot] per user quota in mysql

2013-06-10 Thread John Doe
Hi,
Thank you for responding but i have read in the dovecot documentation and
in postfixadmin documentation that:

 - The setup gets userdb and passdb info from MySQL as well as quotas (
postfixadmin documentation )
-  Quota backend specifies the method how Dovecot keeps track of the
current quota usage. They don't (usually) specify users' quota limits,
that's done by returning extra fields from userdb (
http://wiki2.dovecot.org/Quota )

My question is that if i can keep per user quota limits in a mysql database
and user ldap database for authentication.
I can't see anywhere in the docs if i can return user and password data
from ldap and per user quota from mysql, if i am mistaken, please give me
the link.

Thank you!


On Sun, Jun 9, 2013 at 7:22 AM, Benny Pedersen m...@junc.eu wrote:

 John Doe skrev den 2013-06-07 15:32:


  I have set up dovecot to use ldap authentication and it works great.
 I wonder if it is possible to use mysql for user quota and still keep my
 ldap authentication.


 yes no problem, for more help check postfixadmin on sf.net, even if you
 dont use it, there is good examples on configure it with dovecot quotas

 --
 senders that put my email into body content will deliver it to my own
 trashcan, so if you like to get reply, dont do it



[Dovecot] dovecot segfaults after upgrade

2013-06-10 Thread Thomas Blomenkamp


Using dovecot on debian oldstable (squeeze) with daily builded repository, 
after an upgrade this morning, dovecot always shows the following error:


2013 Jun 10 11:07:22 mailstore imap(tblomenk): Fatal: master: 
service(imap): child 3016 killed with signal 11 (core dumps disabled)
Jun 10 11:07:22 mailstore kernel: [ 1589.400741] imap[3016]: segfault at 
7fffd9048ff8 ip 7f91417e2c3b sp 7fffd9049000 error 6 in 
libdovecot.so.0.0.0[7f9141796000+bc000]


Please notice: I can reproduce the error by trying to access my mailbox 
with alpine or mutt, while alpine only can access my inbox. If I try to 
access my other folders, the error happens. Mutt always closes the 
connection after trying to login. It is a little bit strange because with 
a newer client like roundcube I can access all my folders successfully.


In all kernel error messages it seems that

error 6 in libdovecot.so.0.0.0[7f9141796000+bc000]

is always the same.

Also notice that dovecot is still running, it is just that you cannot 
access other folders than your inbox with pine.


mailstore:~# dovecot --version
2.2.2 (45399357008a)

mailstore:~# dovecot -n
# 2.2.2 (45399357008a): /etc/dovecot/dovecot.conf
doveconf: Warning: service auth { client_limit=1000 } is lower than 
required under max. load (1324)
doveconf: Warning: service anvil { client_limit=1000 } is lower than 
required under max. load (1227)

# OS: Linux 2.6.32-5-amd64 x86_64 Debian 6.0.7
auth_debug = yes
auth_gssapi_hostname = mailstore.math.uni-bielefeld.de
auth_mechanisms = plain login
auth_username_format = %n
auth_verbose = yes
lmtp_save_to_detail_mailbox = yes
log_timestamp = %Y-%m-%d %H:%M:%S 
login_greeting = Hi, the IMAP Mail-System on 
mailstore.math.uni-bielefeld.de is ready.

login_log_format_elements = user=%u method=%m rip=%r lip=%l %c
mail_location = mdbox:~/mdbox
mail_plugins = acl
mail_privileged_group = mail
managesieve_notify_capability = mailto
managesieve_sieve_capability = fileinto reject envelope encoded-character 
vacation subaddress comparator-i;ascii-numeric relational regex imap4flags 
copy include variables body enotify environment mailbox date ihave 
spamtest spamtestplus imapflags notify

namespace {
  list = children
  location = mdbox:%%h/mdbox
  prefix = shared/%%u/
  separator = /
  subscriptions = no
  type = shared
}
namespace inbox {
  inbox = yes
  location =
  mailbox Drafts {
auto = subscribe
special_use = \Drafts
  }
  mailbox Sent {
auto = subscribe
special_use = \Sent
  }
  mailbox Spam {
auto = subscribe
special_use = \Junk
  }
  mailbox Trash {
auto = subscribe
special_use = \Trash
  }
  prefix =
  separator = /
  type = private
}
passdb {
  args = /etc/dovecot/dovecot-ldap.conf
  driver = ldap
}
plugin {
  acl = vfile
  acl_shared_dict = file:/var/lib/dovecot/shared-mailboxes.db
  mail_log_events = delete undelete expunge copy mailbox_delete 
mailbox_rename

  mail_log_fields = uid box msgid size
  mail_log_group_events = no
  quota = maildir:User quota
  quota_rule = *:storage=0
  sieve = ~/.dovecot.sieve
  sieve_dir = ~/sieve
  sieve_extensions = +imapflags +notify +spamtest +spamtestplus
  sieve_spamtest_max_value = 100
  sieve_spamtest_status_header = X-PMX-Spam: 
\.*Probability=([[:digit:]]+)%+\.*

  sieve_spamtest_status_type = score
}
protocols = imap lmtp sieve
service auth {
  unix_listener auth-master {
mode = 0600
user = vmail
  }
  unix_listener auth-userdb {
user = vmail
  }
  user = root
}
service imap-login {
  inet_listener imap {
port = 143
  }
  inet_listener imaps {
port = 0
  }
  process_limit = 1024
  process_min_avail = 10
  service_count = 1
}
service imap {
  process_limit = 1024
}
service lmtp {
  unix_listener /var/spool/postfix/private/dovecot-lmtp {
group = postfix
mode = 0660
user = postfix
  }
}
service managesieve-login {
  inet_listener sieve {
port = 4190
  }
  service_count = 1
  vsz_limit = 64 M
}
ssl_cert = /etc/dovecot/mailstore_cert_chain.pem
ssl_key = /etc/ssl/private/mcl_key.pem
userdb {
  args = uid=vmail gid=vmail home=/var/vmail_san2/dovecot/%1Lu/%Lu
  driver = static
}
protocol imap {
  mail_plugins = acl imap_acl
}
protocol lmtp {
  mail_plugins = acl sieve
}
protocol sieve {
  mail_max_userip_connections = 50
  managesieve_implementation_string = Dovecot Pigeonhole
  managesieve_logout_format = bytes=%i/%o
  managesieve_max_compile_errors = 5
  managesieve_max_line_length = 65536
}

Regards,

Thomas


[Dovecot] Mailbox conversion/importing

2013-06-10 Thread Alan Brown


I've been tasked with importing a large bunch of mbox folders (about 
500) into an existing mdbox setup in Dovecot 2.1


As far as I can see, dsync mirror or backup are both inappropriate 
ways of doing this. Does anyone have any suggestions about how I could 
proceed?


Thanks in advance






Re: [Dovecot] Wait for interface to become available instead of dying?

2013-06-10 Thread Reindl Harald

Am 10.06.2013 11:45, schrieb Sebastian Arcus:
 At the moment, if one of the interfaces specified with listen= in 
 dovecot.conf is not up when Dovecot is started,
 then Dovecot just refuses to start. Is there an option to make Dovecot start 
 anyway, and just use the interface
 when it becomes available? It is inconvenient to have Dovecot refuse to start 
 during boot because some interface is
 temporarily not available.
 
 Then again, maybe there is some strong security reasoning behind the way 
 Dovecot behaves at the moment?

the main question is why do you not order the start of your services correctly
how should a application bind to a specific interface if it is not up?

listening on * is no problem in this case but you can hardly bind
to a non existing interface



signature.asc
Description: OpenPGP digital signature


Re: [Dovecot] Mailbox conversion/importing - SOLVED

2013-06-10 Thread Alan Brown

On 10/06/13 12:03, Alan Brown wrote:


I've been tasked with importing a large bunch of mbox folders (about
500) into an existing mdbox setup in Dovecot 2.1

As far as I can see, dsync mirror or backup are both inappropriate
ways of doing this. Does anyone have any suggestions about how I could
proceed?


I've finally discovered doveadm import.

In this instance:  doveadm -Dv import -u [user]  mbox:/full/path/to/mbox 
old-mbox all



Relative paths don't work. :)






Re: [Dovecot] Mailbox conversion/importing

2013-06-10 Thread Robert Brockway

On Mon, 10 Jun 2013, Alan Brown wrote:

I've been tasked with importing a large bunch of mbox folders (about 500) 
into an existing mdbox setup in Dovecot 2.1


As far as I can see, dsync mirror or backup are both inappropriate ways 
of doing this. Does anyone have any suggestions about how I could proceed?


I've done a variety of mail migrations over the years, including some that 
were quite large (hundreds of thousands of accounts).  I looked at a few 
options when doing the first one and ended up concluding that pop/imap was 
the best way to go.  A specialised migration tool must be less tested (and 
perhaps more buggy) than pop/imap servers that are in use around the world 
constantly.  By using pop/imap proxies we were able to do migrations that 
were completely transparent to users.


This presumes that you are migrating from one server to another.  I've 
always done it like this, rather than having to worry about multiple 
storage formats on the one server.


Cheers,

Rob

--
Email: rob...@timetraveller.org Linux counter ID #16440
IRC: Solver (OFTC  Freenode)
Web: http://www.practicalsysadmin.com
Director, Software in the Public Interest (http://spi-inc.org/)
Information is a gas


Re: [Dovecot] Mailbox conversion/importing

2013-06-10 Thread Frerich Raabe
Hi Robert,

On Jun 10, 2013, at 8:23 AM, Robert Brockway rob...@timetraveller.org wrote:
 I've done a variety of mail migrations over the years, including some that 
 were quite large (hundreds of thousands of accounts).  I looked at a few 
 options when doing the first one and ended up concluding that pop/imap was 
 the best way to go.  A specialised migration tool must be less tested (and 
 perhaps more buggy) than pop/imap servers that are in use around the world 
 constantly.  By using pop/imap proxies we were able to do migrations that 
 were completely transparent to users.

I think this sounds very plausible - can you maybe elaborate a bit on how you 
did this exactly?

Would you say that it even makes sense to use a proxy-based migration if you're 
moving from one Dovecot installation (serving just IMAP) to another?

I'm just asking because I'm planning to replace a FreeBSD-based Dovecot setup 
(serving just IMAP) to Debian. I already have the Debian system set up, but I'm 
still undecided how to do the move in a way which is a) preferrably transparent 
to users and b) possibly even allows me to quickly switch back to the old 
system again, just in case.

-- 
Frerich Raabe - ra...@froglogic.com
www.froglogic.com - Multi-Platform GUI Testing








Re: [Dovecot] Wait for interface to become available instead of dying?

2013-06-10 Thread Jeroen Massar
On 2013-06-10 02:45, Sebastian Arcus wrote:
 At the moment, if one of the interfaces specified with listen= in
 dovecot.conf is not up when Dovecot is started, then Dovecot just
 refuses to start. Is there an option to make Dovecot start anyway, and
 just use the interface when it becomes available? It is inconvenient to
 have Dovecot refuse to start during boot because some interface is
 temporarily not available.
 
 Then again, maybe there is some strong security reasoning behind the way
 Dovecot behaves at the moment?

Depending on platform, but on Linux:

sysctl -w net.ipv4.ip_nonlocal_bind = 1

And presto. Do note that figuring out that some applications are then
misconfigured is a lot of fun, though 'netstat -anp' will help with
that. (-p only as root on again Linuxes)

Greets,
 Jeroen



Re: [Dovecot] Wait for interface to become available instead of dying?

2013-06-10 Thread Sebastian Arcus

On 10/06/13 13:14, Reindl Harald wrote:


Am 10.06.2013 11:45, schrieb Sebastian Arcus:

At the moment, if one of the interfaces specified with listen= in 
dovecot.conf is not up when Dovecot is started,
then Dovecot just refuses to start. Is there an option to make Dovecot start 
anyway, and just use the interface
when it becomes available? It is inconvenient to have Dovecot refuse to start 
during boot because some interface is
temporarily not available.

Then again, maybe there is some strong security reasoning behind the way 
Dovecot behaves at the moment?


the main question is why do you not order the start of your services correctly
how should a application bind to a specific interface if it is not up?


The order of services is fine as it is. The problem is that if any of 
the interfaces Dovecot is supposed to be binding to is missing, Dovecot 
seems to refuse to start at all - instead of just binding to what is 
available. The openvpn service for example might have been reconfigured 
on a different IP. On next reboot, there is no imap server available for 
any interface. One of the network cards might go faulty. On next reboot 
- not imap server.


Exim seems to be happy to start regardless of what is available - but 
I'm not sure of the intricacies of how they do it.




[Dovecot] IMAP

2013-06-10 Thread Jason Lock
We are using version 1.2.17 and recently are experiencing major issues with 
performance, which we believe have isolated to IMAP sessions.

We have 3 servers running Dovecot, with a central store shared via NFS.  Things 
have been running quite well for months now, with the latest issues appearing 
within the last week.

As an experiment have 2 of the server running and only accepting POP3 
connecitons no IMAP, and the 3rd server only accepting IMAP connections and no 
POP3.

When the issue occurred today, stopping dovecot on the IMAP only server allowed 
POP3 to resume to normal operations 5-10 minutes later.  Leaving IMAP disabled 
for a period of time (about 30 mins) and then re-enabling seemed to worked the 
first time.  Subsequent times, the issue appeared shortly after re-enabling 
IMAP.

Our webmail solution connects via IMAP, so when disabled this also impact 
clients using the webmail.

Running only POP3 while IMAP is disabled we do not appear to have any issues.

At this point, looking for any advice.  We believe the number of devices 
utilizing IMAP has increased significantly for us, and whether or not a 
specific device is the cause we have not been able to determine.

Anyone else experiencing a similar problem that appears related to IMAP?



Re: [Dovecot] IMAP

2013-06-10 Thread Rick Romero

 Quoting Jason Lock jl...@csolve.net:


We are using version 1.2.17 and recently are experiencing major issues
with performance, which we believe have isolated to IMAP sessions.

We have 3 servers running Dovecot, with a central store shared via NFS. 
Things have been running quite well for months now, with the latest
issues appearing within the last week.

As an experiment have 2 of the server running and only accepting POP3
connecitons no IMAP, and the 3rd server only accepting IMAP connections
and no POP3.

When the issue occurred today, stopping dovecot on the IMAP only server
allowed POP3 to resume to normal operations 5-10 minutes later.  Leaving
IMAP disabled for a period of time (about 30 mins) and then re-enabling
seemed to worked the first time.  Subsequent times, the issue appeared
shortly after re-enabling IMAP.

Our webmail solution connects via IMAP, so when disabled this also
impact clients using the webmail.

Running only POP3 while IMAP is disabled we do not appear to have any
issues.

At this point, looking for any advice.  We believe the number of devices
utilizing IMAP has increased significantly for us, and whether or not a
specific device is the cause we have not been able to determine.
Anyone else experiencing a similar problem that appears related to IMAP?


I'd suggest checking your I/O load on the NFS server.    Especially If
you have a need for that many front-end machines, that's the first place
I'd check.  I'm not sure if Dovecot 1.x support it offhand, but moving the
indexes to a local SSD on the IMAP only server might help if it's an I/O
issue..
You could also try running IMAP on the NFS server - assuming it's capable.

Personally, I had a hell of a time with certain versions of FreeBSD as an
NFS Client. In my case the server that acted up was an SMTP server for
incoming delivery only. For whatever reason, it would cause EXTREMELY slow
access on the NFS host and drag everything down.   This happened to me
with both FreeBSD and OpenSolaris as the NFS host. I have no idea why, but
an upgrade/rebuild fixed my issues. This probably isn't your issue, my
problems cropped up almost immediately, but it's something to keep in
mind.  Check your load on the NFS server first.

Rick


[Dovecot] dovecot corrupted transaction log

2013-06-10 Thread John Fawcett
Hi I came across this error which happend immedately after a mail
delivery to the inbox. Should I look for the problem externally to
dovecot (ie. file system, operating system) or within dovecot? I never
saw this error before installing 2.2.1, with 2.2.2 I seemed to get even
more of them so currently back on 2.2.1

Thanks
John


Jun 11 00:00:05 rosalia dovecot: imap(myemail@mydomain): Error:
Corrupted transaction log file
/var/vmail/mydomain/myemail@mydomain/dovecot.index.log seq 311: file
size shrank (1184  1304) (sync_offset=1304)
Jun 11 00:00:05 rosalia dovecot: imap(myemail@mydomain): Error: Index
/var/vmail/mydomain/myemail@mydomain/dovecot.index: Lost log for seq=310
offset=32816
Jun 11 00:00:05 rosalia dovecot: imap(myemail@mydomain): Warning:
fscking index file /var/vmail/mydomain/myemail@mydomain/dovecot.index
Jun 11 00:00:05 rosalia dovecot: imap(myemail@mydomain): Error: Index
/var/vmail/mydomain/myemail@mydomain/dovecot.index: Lost log for seq=310
offset=32816
Jun 11 00:00:05 rosalia dovecot: imap(myemail@mydomain): Warning:
fscking index file /var/vmail/mydomain/myemail@mydomain/dovecot.index
Jun 11 00:00:05 rosalia dovecot: imap(myemail@mydomain): Error:
/var/vmail/mydomain/myemail@mydomain/dovecot.index log position went
backwards (310,32816  311,1304)
Jun 11 00:00:05 rosalia dovecot: imap(myemail@mydomain): Error: Index
/var/vmail/mydomain/myemail@mydomain/dovecot.index: Lost log for seq=310
offset=32816
Jun 11 00:00:05 rosalia dovecot: imap(myemail@mydomain): Warning:
fscking index file /var/vmail/mydomain/myemail@mydomain/dovecot.index
Jun 11 00:00:05 rosalia dovecot: imap(myemail@mydomain): Disconnected:
Internal error occurred. Refer to server log for more information.
[2013-06-11 00:00:05] in=5596 out=27722
Jun 11 00:02:11 rosalia dovecot: imap-login: Login:
user=myemail@mydomain, method=PLAIN, rip=81.174.4.175,
lip=80.237.194.64, mpid=5824, TLS, session=lGC47tPexABRrgSv
Jun 11 00:02:11 rosalia dovecot: imap(myemail@mydomain): Error:
Transaction log file
/var/vmail/mydomain/myemail@mydomain/dovecot.index.log: marked corrupted



Re: [Dovecot] Wait for interface to become available instead of dying?

2013-06-10 Thread Reindl Harald

Am 10.06.2013 21:04, schrieb Sebastian Arcus:
 On 10/06/13 13:14, Reindl Harald wrote:

 Am 10.06.2013 11:45, schrieb Sebastian Arcus:
 At the moment, if one of the interfaces specified with listen= in 
 dovecot.conf is not up when Dovecot is started,
 then Dovecot just refuses to start. Is there an option to make Dovecot 
 start anyway, and just use the interface
 when it becomes available? It is inconvenient to have Dovecot refuse to 
 start during boot because some interface is
 temporarily not available.

 Then again, maybe there is some strong security reasoning behind the way 
 Dovecot behaves at the moment?

 the main question is why do you not order the start of your services 
 correctly
 how should a application bind to a specific interface if it is not up?
 
 The order of services is fine as it is. The problem is that if any of the 
 interfaces Dovecot is supposed to be
 binding to is missing, Dovecot seems to refuse to start at all

where i work and config servers *i want* the to fail if the config is wrong

 instead of just binding to what is available

is not a predictable configuration if you specify ecplicit interfaces

 openvpn service for example might have been reconfigured on a different IP

so why the hell to you not config dovecot with address = * if you want this

 On next reboot, there is no imap server available for any interface

which is good because you recognize something goes wrong
and if you want it to listen to whatis available avoid
configs with specific interfaces

 One of the network cards might go faulty. On next reboot - not imap server.

so what - if hardware dies you normally want to know it instead
hav eit somehow masqueraded

 Exim seems to be happy to start regardless of what is available

dovecot too as any other service if you configure it not explicitly for 
specific interfaces



signature.asc
Description: OpenPGP digital signature