- dovecot - dovecot.org

pop3-login segfaulting
by James Shewey 16 Oct '14

16 Oct '14

I upgraded dovecot from v1 to v2 as part of a CentOS upgrade (5.6->6.5) and am now running on 2.0.9. Any time I connect to the pop or imap ports however I get "master: Error: service(imap-login): child 2860 killed with signal 11 (core not dumped - add -D parameter to service imap-login { executable }" after following the core dump instructions at http://www.dovecot.org/bugreport.html. The problem is that I'm not sure how these processes are spawned in order to add the switch. Can I spawn them directly? If so, I ought to be able to do an strace to track down the issue as well. Regards, James Shewey NOC Systems Administrator I Product Support Engineering [Description: Panasonic Avionics] Panasonic Avionics Corporation Phone: +1 (949) 672-2000 Direct: +1 (949) 595-5187

2 2

Permissions problem on new installation
by Cliff Hayes 15 Oct '14

15 Oct '14

I am setting up a new mail server using imap only on Scientific Linux 6.5 and dovecot 2.0.9. How do I fix this problem? I keep getting these errors regardless of permissions ... even up to 777. Oct 15 13:56:01 avalon dovecot: imap(chayes(a)domain.tld): Error: opendir(/var/email/chayes(a)domain.tld/Maildir) failed: Permission denied (euid=497(chayes) egid=12(mail) missing +r perm: /var/email/chayes(a)domain.tld/Maildir, euid is not dir owner) Oct 15 13:56:01 avalon dovecot: imap(chayes(a)domain.tld): Error: opendir(/var/email/chayes(a)domain.tld/Maildir) failed: Permission denied (euid=497(chayes) egid=12(mail) missing +r perm: /var/email/chayes(a)domain.tld/Maildir, euid is not dir owner) Oct 15 13:56:01 avalon dovecot: imap(chayes(a)domain.tld): Error: opendir(/var/email/chayes(a)domain.tld/Maildir/new) failed: Permission denied (euid=497(chayes) egid=12(mail) missing +r perm: /var/email/chayes(a)domain.tld/Maildir/new, euid is not dir owner) Oct 15 13:56:07 avalon dovecot: imap(chayes(a)domain.tld): Error: opendir(/var/email/chayes(a)domain.tld/Maildir/new) failed: Permission denied (euid=497(chayes) egid=12(mail) missing +r perm: /var/email/chayes(a)domain.tld/Maildir/new, euid is not dir owner) Yet when i check, chayes is directory owner as follows (currently showing 760 but I've tried everything): drwx------. 3 chayes mail 4096 Oct 15 12:56 chayes(a)domain.tld drwxrwx---. 6 chayes mail 4096 Oct 15 13:55 Maildir drwxrw----. 2 chayes mail 4096 Oct 15 12:56 cur -rw-------. 1 chayes mail 104 Oct 15 12:56 dovecot.index.log -rw-------. 1 chayes mail 0 Oct 15 12:56 dovecot.mailbox.log -rw-------. 1 chayes mail 51 Oct 15 12:56 dovecot-uidlist drwxrw----. 2 chayes mail 4096 Oct 15 12:56 new -rw-------. 1 chayes mail 6 Oct 15 12:56 subscriptions drwxrw----. 2 chayes mail 4096 Oct 15 12:56 tmp dovecot config: protocols = imap mail_access_groups=mail mail_location = maildir:~/Maildir Thanks in advance.

1 1

Re: delete/archive old mail
by Joseph Tam 15 Oct '14

15 Oct '14

"F." <sottilette(a)rfx.it> writes: >>> doveadm-altmove seems interesting, but in the man page it is reported >>> only work with dbox (dbox-only), and my inboxes are in mbox format. >>> >>> Any other advice? > > As I originally wrote, my need is script or a command like the old > Expire_mail.pl that parse a local mbox folder and move/copy mail older > than NN days to another local mbox folder. > Probably for a Perl programmer is only matter of modify this script or > something similar. > I searched in docs, wiki, ... but not noticied something that can simply > be inserted in a cron job to do the work. > > Only migration tools or complex (for me), library that can be used to > manipulate folders :-) I don't understand how you could have found doveadm-altmove, but miss doveadm-move -- doesn't this dovecot command do exactly what you want? $ man doveadm-move NAME doveadm-move - Move messages matching the given search query into another mailbox ... EXAMPLE Move janes messages - received in September 2011 - from her INBOX into her archive. doveadm move -u jane Archive/2011/09 mailbox 2011-10-01 SINCE 01-Sep-2011 [Is this a typo: "2011-10-01" should be "INBOX"?] The example is fairly close to what you want. Apart from the obvious change to user and mailbox name, the condition would be "savedbefore ${NN}d". You might have to follow this up with an expunge if this operation does not do that. Joseph Tam <jtam.home(a)gmail.com>

2 1

mmap_disable=yes not honored always
by Peter Benko 15 Oct '14

15 Oct '14

Hi all, I'm experimenting with having the mail store on a 9p file system that lacks mmap() functionality. So I disabled it in dovecot: mmap_disable = yes However, I keep getting the following error messages in my log: Oct 15 16:55:00 computer-name dovecot: imap user(a)domain.com[192.168.1.3] Error: mmap() failed with file /mail/mailboxes/domain.com/user/indexpvt-accounts/other-user/mailboxes/INBOX/dovecot.index.pvt.log: Invalid argument I only get those messages for the shared folder's private indexes. This setup works perfectly on an ext2/3/4 fs. Any opinions, is this a bug? Is mmap_disable not mandatory for all operations? Best regards, Peter

1 0

Request to list owner
by Jorge Bastos 15 Oct '14

15 Oct '14

Hi list owner, Sorry for this. Is it possible to change the reply-to in mailman to be the list, and not the person who replied? Also the [Dovecot] on the subject would be handy. Just two things to get better. Thanks,

10 15

permissions when moving mails
by Alexander 'Leo' Bergolth 15 Oct '14

15 Oct '14

Hi! I am using Unix groups to control access to shared mailboxes. This works fine as long as files are not moved between mailboxes that are owned by different groups. The move operation doesn't change group ownership of the mail. Is there a way to force dovecot to e.g. inherit the ownership from the parent folder, not only for new messages bot also for messages moved from one folder to another? The wiki says: http://wiki2.dovecot.org/SharedMailboxes/Permissions -------------------- 8< -------------------- When mails are copied between Maildirs, it's usually done by hard linking. If the source and destination directory permissions are different, Dovecot create a new file and copies data the slow way so that it can assign the wanted destination permissions. The source and destination permission lookups are done only by looking at the mailbox root directories' permissions, not individual mail files. This may become a problem if the mail files' permissions aren't as Dovecot expects. -------------------- 8< -------------------- Does dovecot only look at the permissions or does it also take user/group ownership into account? The behavior I experience is that even though the mailbox directories are owned by different groups, dovecot doesn't copy the mails. (See my example below.) Setting maildir_copy_with_hardlinks = no seems to fix the problem but maybe this behavior should be considered as a bug? I am using dovecot 2.2.14 and maildir mailboxes. Cheers, --leo P.S.: Here is an example of the current behavior: before the move from PAS to GSD: -------------------- 8< -------------------- GSD/.Eingang/cur/: total 0 drwxrws--- 2 nobody gsd 6 Oct 13 13:46 . drwxrws--- 5 nobody gsd 81 Oct 15 09:41 .. PAS/.Eingang/cur/: total 4 drwxrws--- 2 nobody pas 89 Oct 15 09:39 . drwxrws--- 5 nobody pas 79 Oct 15 09:39 .. -rw-rw-r-- 1 nobody pas 4073 Jul 8 08:14 1404967500.M656935P20690.samba.intern.rk-klosterneuburg.at,S=4073,W=4213:2,S -------------------- 8< -------------------- after the move: -------------------- 8< -------------------- GSD/.Eingang/cur/: total 4 drwxrws--- 2 nobody gsd 87 Oct 15 09:42 . drwxrws--- 5 nobody gsd 81 Oct 15 09:42 .. -rw-rw-r-- 1 nobody pas 4073 Jul 8 08:14 1413358925.M22574P3008.samba.intern.rk-klosterneuburg.at,S=4073,W=4213:2,S PAS/.Eingang/cur/: total 0 drwxrws--- 2 nobody pas 6 Oct 15 09:42 . drwxrws--- 5 nobody pas 79 Oct 15 09:42 .. -------------------- 8< -------------------- Mailbox directories are: -------------------- 8< -------------------- drwxrws--- 4 nobody pas 143 Nov 25 2013 PAS drwxrws--- 5 nobody pas 79 Oct 15 09:57 PAS/.Eingang drwxrws--- 10 nobody gsd 4096 Dec 11 2013 GSD drwxrws--- 5 nobody gsd 81 Oct 15 09:42 GSD/.Eingang -------------------- 8< -------------------- (In my own litte world, the mail should now be owned by the gsd group.. :-)) -- e-mail ::: Leo.Bergolth (at) wu.ac.at fax ::: +43-1-31336-906050 location ::: IT-Services | Vienna University of Economics | Austria

1 0

Dbox and Exim
by C Peters 15 Oct '14

15 Oct '14

The Dovecot wiki [1] doesn't list either Exim, Postfix or anything else as supporting the dbox. I did some searching in the Exim archives and have found only one message mentioning dbox, sdbox or mdbox. [2] I think an RFC would go a long way towards getting the format supported. Has Timo, or anyone else, considered submitting an RFC to the IETF? My question was how are we going to use it if Exim doesn't support it. It looks like we can filter it and convert it with dovecot-lda. [3] I have generally avoided using Procmail for a few years because of various problems and it's a dead software project that hasn't been developed in what 15 to 20 years. I am thinking about trying it with Exim's built in Maildrop filtering support. Has anybody tried that? What else are people using to enable dbox? The more specific question on the problem I am trying to solve. I have been using mbox for years, and I can't even get Dovecot, IMAP and Thunderbird working on the LAN because it can't handle the GB's of old mail. Mutt works on the local machine, but I want something that works while I am using something other than ssh and mutt. For example I want to try Mailpile and K9. I would like to try dbox, but given it's lack of support in Exim or Postfix, I'll probably convert to maildir. Thanks, Chuck 1. http://wiki2.dovecot.org/MailboxFormat 1. https://lists.exim.org/lurker/message/20110220.014144.87010abd.en.html Thread view: https://lists.exim.org/lurker/thread/20110216.223226.015539ba.en.html#i2011… 2. http://ramblings.narrabilis.com/mdboxprocmail

7 9

Proxy problem: "[COMPRESSIONACTIVE] TLS compression already enabled"
by Ralf Hildebrandt 15 Oct '14

15 Oct '14

We're proxying using 2.2.14~rc1 (on our IMAP Proxy) to two dovecot backend servers running dovecot-2.2.13-r1 When we're using Thundebird to connect to the dovecot proxy, we're getting the message "The mail server for account ACCOUNTNAME responded: [COMPRESSIONACTIVE] TLS compression already enabled" But why? Capabilities reported by the proxy: . OK [CAPABILITY IMAP4rev1 LITERAL+ SASL-IR LOGIN-REFERRALS ID ENABLE IDLE SORT SORT=DISPLAY THREAD=REFERENCES THREAD=REFS THREAD=ORDEREDSUBJECT MULTIAPPEND URL-PARTIAL CATENATE UNSELECT CHILDREN NAMESPACE UIDPLUS LIST-EXTENDED I18NLEVEL=1 CONDSTORE QRESYNC ESEARCH ESORT SEARCHRES WITHIN CONTEXT=SEARCH LIST-STATUS SPECIAL-USE BINARY MOVE SEARCH=FUZZY COMPRESS=DEFLATE QUOTA] Logged in Capabilities of the backend servers when connecting directly: . OK [CAPABILITY IMAP4rev1 LITERAL+ SASL-IR LOGIN-REFERRALS ID ENABLE IDLE SORT SORT=DISPLAY THREAD=REFERENCES THREAD=REFS THREAD=ORDEREDSUBJECT MULTIAPPEND URL-PARTIAL CATENATE UNSELECT CHILDREN NAMESPACE UIDPLUS LIST-EXTENDED I18NLEVEL=1 CONDSTORE QRESYNC ESEARCH ESORT SEARCHRES WITHIN CONTEXT=SEARCH LIST-STATUS SPECIAL-USE BINARY MOVE SEARCH=FUZZY COMPRESS=DEFLATE QUOTA] Logged in -- [*] sys4 AG http://sys4.de, +49 (89) 30 90 46 64 Franziskanerstraße 15, 81669 München Sitz der Gesellschaft: München, Amtsgericht München: HRB 199263 Vorstand: Patrick Ben Koetter, Marc Schiffbauer Aufsichtsratsvorsitzender: Florian Kirstein

3 5

bandwidth shaping
by Alan Bailward 14 Oct '14

14 Oct '14

Hey all, sorry if this has been asked before, but I haven’t really found a good solution. I have a mac at home and randomly, maybe once a week or once a month, Mail.app will go crazy and use as much bandwidth as I have to pull down mail. I’m not home so I can’t tell what it’s done or why, and sometimes it stops after an hour, sometimes it goes all night, leaving me with a potentially big bandwidth bill for my colo’d server which hosts my mail (dovecot 1.2.9, postfix 2.7.0 on Ubuntu LTS 10.04). I’m wondering if there’s a way to restrict bandwidth that’s the “accepted” and proper way with dovecot? There isn’t anything (AFAICT) built into dovecot, and the alternative seems to be setting up bandwidth shaping with ‘ip’, but that is a pretty horrible nest of ugliness. Is there a fairly simple setup to set it so imap/pop traffic can only ever use 1mb/s (or some arbitrary number)? I’m not looking for a one liner, but each time this happens I say “I’m going to figure this out and fix this”, then get lost in ip and shaping queues and give up in frustration until the next time it happens, when I say “I’m *really* going to figure this out this time…”. Anyone have any pointers? TIA Alan

2 1

auth-worker goes wild
by Hans Morten Kind 14 Oct '14

14 Oct '14

dovecot-2.2.13 running on Solaris-10, app 5000 concurrent connections, and a pretty standard config where passdb is set up with driver = pam. We have had some situations where an auth-worker gets into a loop, the load on the server rises and the auth-master seems blocked untill finally no new imap nor pop session are accepted. A 'doveadm reload' restores normal operations. With debug logging enabled we are observing this: auth: Debug: pam(,10.10.30.33,<7WMu3UkEGQCBsR4h>): cache miss auth-worker(24556): Debug: pam(,10.10.30.33): lookup service=dovecot auth-worker(24556): Debug: pam(,10.10.30.33): #1/1 style=2 msg=Please enter user name: ... auth-worker(24556): Debug: pam(,10.10.30.33): #1/1 style=2 msg=Please enter user name: etc etc After some time dovecot only logs Warning: Auth process not responding, delayed sending initial response (greeting) Disconnected: Auth process broken (disconnected before auth was ready, waited 26 secs) 'truss -wall -aep PID' on the looping auth-worker shows the loop: 0.0757 write(2, 0x080AE0C0, 71) = 71 0101 2 4 5 5 6 p a m ( , 1 0 . 1 0 . 3 0 . 3 3 ) : # 1 / 1 s t y l e = 2 m s g = P l e a s e e n t e r u s e r n a m e : \n 0.0826 write(2, 0x080AE0C0, 71) = 71 0101 2 4 5 5 6 p a m ( , 1 0 . 1 0 . 3 0 . 3 3 ) : # 1 / 1 s t y l e = 2 m s g = P l e a s e e n t e r u s e r n a m e : \n The host at 10.10.30.33 is running RoundCube for app 1000 concurrent users. This problem started after we provided our users RoundCube, and every time we end up in this stuck situation, it is the RoundCube-server making the connections where the username has zero length. I am unable to reproduce how RoundCube puts dovecot into this mess. With the enclosed patch installed, the loop terminates and normal operation are restored immediately: auth: Debug: pam(,10.10.30.33,<Q8kV6AEFYwCBsR4h>): cache miss auth-worker(18658): Debug: pam(,10.10.30.33): lookup service=dovecot auth-worker(18658): Debug: pam(,10.10.30.33): #1/1 style=2 msg=Please enter user name: auth-worker(18658): Fatal: NO USER? auth-worker(18658): Error: Raw backtrace: 0xfef0798c -> 0xfef06d3b -> 0x807ca6e -> 0xfecc460e -> 0xfecc1d1e -> 0xfe3c0af4 -> 0xfecc2608 -> 0xfecc27f1 -> 0x807cc5e -> 0x807d162 -> 0x807d310 -> 0x8069ca1 -> 0x806ab72 -> 0x806aeec -> 0xfef1e866 -> 0xfef2001c -> 0xfef1e9ff -> 0xfef1e956 -> 0xfeeb0172 -> 0x8071bf2 -> 0x805a650 auth: Error: auth worker: Aborted PASSV request for : Worker process died unexpectedly I think however, that the pam driver should not have been initiated when the username has zero length, definitely there is nobody around to answer the prompt for a username ... hmk

2 2