- dovecot - dovecot.org

[Dovecot] ID command and dovecot
by Jernej Porenta 02 May '14

02 May '14

Heya, we are seeing some strange issues with IMAP ID command using latest Roundcube builds on dovecot 2.2.10 responding "Input buffer full, aborting". Roundcube webmail connects to IMAP server and issues ID command first with some long arguments to ID command, however shorter than 1024 octets, which are specified in RFC2971 for ID IMAP extension. The situation is easily reproducible (see http://pastebin.com/q4HX4wqb) # telnet hostname 143 * OK [CAPABILITY IMAP4rev1 LITERAL+ SASL-IR LOGIN-REFERRALS ID ENABLE IDLE AUTH=PLAIN AUTH=LOGIN] Server ready. . ID ("name" "Roundcube" "version" "1.0-git" "php" "5.3.3" "os" "Linux" "command" "<$long_less_than_1024_octets_string>") * BYE Input buffer full, aborting However, this problem does not occur when user is already logged in. RFC2971 is saying "Field strings MUST NOT be longer than 30 octets. Value strings MUST NOT be longer than 1024 octets. Implementations MUST NOT send more than 30 field-value pairs." IMHO, dovecot is handling ID command before successful login successfully only if the length of the whole command is less than 1024 octets (exactly 861 octets is working fine, over that "Input buffer full" occurs). I tried finding the reference in the source code, however I am unable to find where this limit actually occurs (maybe IMAP_ARG_STRING imap_arg->type?) thank you in advance, regards, Jernej

2 1

[Dovecot] Finding memory leaks
by Philipp Brüll 02 May '14

02 May '14

Hello, I try to find a memory leak in a dovecot plugin that I develop. In order to find it, it would be helpful to print the total amount of memory that is currently allocated. This print could than spread over the source code and the memory consumption can be tracked. I've tried i_debug("pool size %u", (unsigned int)pool_alloconly_get_total_alloc_size(system_pool)); But that failed. Does someone has a good advice? I would like to avoid complex solutions like valgrind. Best regards, Philipp

2 1

Re: [Dovecot] LMTP error
by Joseph Tam 02 May '14

02 May '14

IT geek 31 <itgeek31(a)googlemail.com> writes: > I've recently switched to LMTP as I'm now using mdbox. However since > switching, mails sent to the root account do not get delivered due to the > following error: > > May 1 18:20:17 Server1 postfix/lmtp[13019]: CAEE91F851: to=<root(a)test.com>, > relay=mail.test.com[private/dovecot-lmtp], delay=1097, > delays=1096/0.45/0.43/0.46, dsn=4.3.0, status=deferred (host > mail.test.com[private/dovecot-lmtp] > said: 451 4.3.0 <root(a)test.com> Invalid user settings. Refer to server log > for more information. (in reply to RCPT TO command)) This setting: > first_valid_uid = 1010 will need to be changed to "0", or better yet, as others have suggested, alias root to some user with UID within 1010..1020. The dovecot logs that you ought to see will be something like May 1 16:12:11 viol dovecot: lda: Error: user root: Mail access for users with UID 0 not permitted (see first_valid_uid in config file, uid from userdb lookup). Joseph Tam <jtam.home(a)gmail.com>

3 2

[Dovecot] Imap user via proxy with starttls - how to rawlogs on proxy ?
by askforstatus 02 May '14

02 May '14

Hello, Is there any way to log user imap dialog with starttls when going via dovecot acting as imap proxy ? My intention is to log each and every user command similar to rawlogs feature. As per my current imap proxy config pre-login working like a charm but post-login is not possible due as i think the nature of proxy feature. Any idea ? Regards -- View this message in context: http://dovecot.2317879.n4.nabble.com/Imap-user-via-proxy-with-starttls-how-… Sent from the Dovecot mailing list archive at Nabble.com.

2 1

[Dovecot] Not backing up cache files
by Jesus Cea 02 May '14

02 May '14

I am using mdbox for dovecot storage. I wonder if I could skip "dovecot.index.cache" files and, in general, "*.cache" files when doing mail backup. Those files are big and change frequently. What I feel from reading documentation [1][2] is that they could be dropped in the backups, to be recreated on the fly if needed. But I would like expert confirmation. [1] http://wiki2.dovecot.org/IndexFiles [2] http://wiki2.dovecot.org/Design/Indexes/Cache Thanks!. -- Jesús Cea Avión _/_/ _/_/_/ _/_/_/ jcea(a)jcea.es - http://www.jcea.es/ _/_/ _/_/ _/_/ _/_/ _/_/ Twitter: @jcea _/_/ _/_/ _/_/_/_/_/ jabber / xmpp:jcea@jabber.org _/_/ _/_/ _/_/ _/_/ _/_/ "Things are not so easy" _/_/ _/_/ _/_/ _/_/ _/_/ _/_/ "My name is Dump, Core Dump" _/_/_/ _/_/_/ _/_/ _/_/ "El amor es poner tu felicidad en la felicidad de otro" - Leibniz

5 5

[Dovecot] BUG dovecot and nginx
by Tobias Oetiker 02 May '14

02 May '14

we have setup a two level proxy configuration for our zimbra server: [ dovecot 2.2.12 (imap proxy mode) ] V [ nginx (imap proxy mode) ] V [ zimbra imap server] and it does not work ... after tying a login, the connection just hangs and ends after 30 seconds with a timeout. - if I try again rightaway in the same dovecot connection, the login goes though without trouble. - looking at this with tshark I can see that on the second attempt, there is no capabilities query, dovecot just goes ahead and connects. if we drop nginx from the setup all is well also (login works directly as expected) - investigating the traffic differences between the setups with and without nginx, we find that for some reason dovecot does not react to the capabily answer from nginx. while it seems to happily accept the same answer when it is coming from zimbra directly. - the only difference I can see on the wire is that nginx is much quicker (2.5ms) in responding than the zimbra (25ms) server and that the position of the STARTTLS and SASL=PLAIN capabilities in the capability answer differs between the two. any ideas ? cheers tobi -- Tobi Oetiker, OETIKER+PARTNER AG, Aarweg 15 CH-4600 Olten, Switzerland www.oetiker.ch tobi(a)oetiker.ch +41 62 775 9902 *** We are hiring IT staff: www.oetiker.ch/jobs ***

2 1

[Dovecot] Rebuilding message guid in sdbox
by Renaud Allard 02 May '14

02 May '14

Hello, I am running dovecot 2.2.12 Current mail storage is sdbox and I am trying to compress all emails, which works fine, except for a few old mailboxes. If I run the following dsync command "dsync -u flea -o "mail_plugins=zlib" mirror maildir:~/ndbox" It fails with: dsync(flea): Error: Exporting mailbox INBOX failed: Backend doesn't support GUIDs, sync with header hashes instead Doing "doveadm dump flea/dbox/mailboxes/INBOX/dbox-Mails/dovecot.index" reveals that there are very old mails (about 5 years ago) without any guid like that one: RECORD: seq=1, uid=1560, flags=0x00 - ext 1 modseq : 16043 (ab3e000000000000) - ext 3 cache : 396 (8c010000) - cache offset=396 size=20, prev_offset = 0 - guid: So, my question is: is there a way to force dovecot to assign a guid to those messages? Best Regards

2 1

[Dovecot] Dovecot Director and MasterUsers
by Daniel Parthey 02 May '14

02 May '14

Hi, the Dovecot Director determines the backend host in some way by hashing the username: http://wiki2.dovecot.org/Director For normal logins username(a)example.org, the director always gets the same hash for the same username and ensures that the login is always proxied to the same backend. But what about MasterUsers in combination with Dovecot Director? http://wiki2.dovecot.org/Authentication/MasterUsers Which configuration directives should be used to make sure that logins username@example.org*masteruser1@example.org username@example.org*masteruser2@example.org username@example.org*masteruser3@example.org go to the same mailbox backend, in order to avoid NFS caching conflicts for mailbox username(a)example.org which should always reside on the same NFS client? Is the master user cut off from behind the master_user_separator? How is the director hash exactly calculated? Can the director hashing algorithm be configured in some way? Regards Daniel

2 1

[Dovecot] BUG: Authentication client sent unknown handshake command
by Jean-Yves Moulin 02 May '14

02 May '14

Hi everybody, I'm using dovecot-2.2.11 on NetBSD. After a kernel upgrade, while using the same dovecot binaries (they were working perfectly until today), I'm experiencing this error (more log at the end of my email): BUG: Authentication client sent unknown handshake command: REQUEST?227672066?18807?2?c717631d60216b2e9ec57a2fa69674b8?session_pid=16343?... I tried to build a newer version of dovecot (2.2.12) but same issue. I know this should come from my system and not dovecot, but how can I debug this kind of problem ? I already enabled various debug (auth_verbose, auth_debug, mail_debug) but I need more information. This issue have already been discussed on dovecot mailing list without solution :-) Thank you very much. Best, jym Log: Apr 16 16:29:16 dodo dovecot: imap-login: Debug: SSL: where=0x2002, ret=1: SSL negotiation finished successfully [::1] Apr 16 16:29:17 dodo dovecot: auth: Debug: client in: AUTH 2 PLAIN service=imap secured session=cE/PvSn3qgAAAAAAAAAAAAAAAAAAAAAB lip=::1 rip=::1 lport=993 rport=65450 resp=<hidden> Apr 16 16:29:17 dodo dovecot: auth: Debug: cache(mylogin@myhost.org,::1,<cE/PvSn3qgAAAAAAAAAAAAAAAAAAAAAB>): hit: <hidden> Apr 16 16:29:17 dodo dovecot: auth: Debug: client passdb out: OK 2 user=mylogin(a)myhost.org Apr 16 16:29:17 dodo dovecot: auth: Error: BUG: Authentication client sent unknown handshake command: REQUEST?227672066?18807?2?c717631d60216b2e9ec57a2fa69674b8?session_pid=16343?... Apr 16 16:29:17 dodo dovecot: imap: Error: Authentication server didn't send valid SPID as expected: MECH PLAIN plaintext Apr 16 16:29:17 dodo dovecot: imap: Error: Disconnected from auth server, aborting (client-pid=18807 client-id=2) Apr 16 16:29:17 dodo dovecot: imap-login: Internal login failure (pid=18807 id=2) (internal failure, 1 successful auths): user=<mylogin(a)myhost.org>, method=PLAIN, rip=::1, lip=::1, mpid=16343, TLS, session=<cE/PvSn3qgAAAAAAAAAAAAAAAAAAAAAB> Apr 16 16:29:17 dodo dovecot: imap-login: Debug: SSL alert: close notify [::1]

2 1

Re: [Dovecot] nfs flush/fsync config settings problem
by Wesley Huang 02 May '14

02 May '14

Hi Alessio, After days sorting the Heartbleed mess, I managed to set up Debian Squeeze amd64 for the test and I'm seeing the same error message: Fatal: nfs flush requires mail_fsync=always # doveconf -n | egrep 'fsync|nfs' mail_fsync = always mail_nfs_index = yes mail_nfs_storage = yes Cheers, Wesley On 14-04-03 09:24 AM, Alessio Cecchi wrote: > Hi, > > I'm using Maildir on Debian 6 64bit, try to "grep" for "fsync" on all > configuration files. > > Try also to install the same dovecot+configuration on Debian 6. > > Let me know! > Ciao > > Il 03/04/2014 15:12, Wesley Huang ha scritto: >> Hi Alessio, >> >> I have 2.2.12 installed in Debian Wheezy amd64 server. mail_fsync is >> specified and detected. Are you using mbox or maildir or dbox? We're >> using mbox and seeing the error. >> >> # dovecot --version >> 2.2.12 >> >> # dovecot -n | grep -E 'sync|nfs' >> mail_fsync = always >> mail_nfs_index = yes >> mail_nfs_storage = yes >> >> >> # egrep 'nfs|fsync' 10-mail.conf | egrep -v '#' >> mail_fsync=always >> mail_nfs_storage = yes >> mail_nfs_index = yes >> >> # uname -r >> 3.2.0-4-amd64 >> >> Cheers, >> >> Wesley >> >> On 14-04-03 04:22 AM, Alessio Cecchi wrote: >>> Il 08/01/2014 09:23, Troton_admin ha scritto: >>>> Hi, >>>> >>>> I need to place a maildir storage on a NFS filesystem, so I set up >>>> the config like this: >>>> >>>> mmap_disable = yes >>>> mail_nfs_storage = yes >>>> mail_nfs_index = yes >>>> mail_fsync = always >>>> lock_method = dotlock (also tried fcntl - local lockd is running, no >>>> success, now using dotlock, to be sure while debugging) >>>> >>>> The problem is that Dovecot (2.2.9/2.2.10 - the relevant code in >>>> mail-index.c is the same) logs a fatal error and quits the session >>>> (no daemon crash/coredump). >>> >>> Hi, >>> >>> I have a similar configuration but with dovecot 2.2.12 and NFS for me >>> works fine. Have you insert the "mail_fsync = always" in the right >>> place? Is there only one "mail_fsync" setting in your config? First >>> check this: >>> >>> root@mx:/etc/dovecot# grep mail_fsync /etc/dovecot/ -R >>> conf.d/10-mail.conf:mail_fsync = always >>> >>> root@mx:/etc/dovecot# dovecot -n | grep -E 'sync|nfs' >>> mail_fsync = always >>> mail_nfs_index = yes >>> mail_nfs_storage = yes >>> maildir_very_dirty_syncs = yes >>> >>> # 2.2.12: /etc/dovecot/dovecot.conf >>> # OS: Linux 2.6.32-5-amd64 x86_64 Debian 6.0.9 >>> >>> Let me know. >>> Ciao >> > >

3 3