August 2010 - dovecot

[Dovecot] MainConfig for dovecot 2.0
by Sascha Peters 31 Aug '10

31 Aug '10

Hey, for dovecot 1.2 there is a very good reference page for all options. http://wiki.dovecot.org/MainConfig is there a reference for dovecot 2.0 in the new wiki? If not, when will be this page online? Thanks a lot. -- Gruß Sascha

2 1

[Dovecot] NTLM authentication does not work
by Alexander Cherniak 31 Aug '10

31 Aug '10

Dear Experts, I configured Samba, Kerberos, etc., can login to my CentOS using ADS account, but missing something in my Dovecot config. Windbind seems to work: [root@aCentOs2 dovecot]# wbinfo -a wAlex%pass plaintext password authentication succeeded challenge/response password authentication succeeded Same is true for ntlm_auth helper: [root@aCentOs2 dovecot]# /usr/bin/ntlm_auth --username=wAlex --password=pass NT_STATUS_OK: Success (0x0) But I couldn't get into Dovecot (with or without auth_use_winbind option): . login wAlex pass . NO [ALERT] Unsupported authentication mechanism. Despite very verbose logging, there no clues there. Bellow's my dovecot -n output: # 2.0.0: /etc/dovecot/dovecot.conf # OS: Linux 2.6.18-194.11.1.el5 i686 CentOS release 5.5 (Final) auth_debug = yes auth_debug_passwords = yes auth_master_user_separator = * auth_mechanisms = ntlm auth_use_winbind = yes auth_verbose = yes auth_verbose_passwords = plain debug_log_path = /var/log/Dovecot/dovecot_debug.log disable_plaintext_auth = no info_log_path = /var/log/Dovecot/dovecot_info.log log_path = /var/log/Dovecot/dovecot.log mail_location = maildir:~/Maildir protocols = imap ssl_cert = </etc/pki/dovecot/certs/dovecot.pem ssl_key = </etc/pki/dovecot/private/dovecot.pem userdb { args = uid=vmail gid=vmail home=/home/vmail/%u allow_all_users=yes driver = static } It might be irrelevant, but although the lines "auth_use_winbind = yes" and "auth_winbind_helper_path = /usr/bin/ntlm_auth" are close together in the config file, I cannot see the latter when run dovecot -n. I am quite new to Dovecot and will appreciate any help. Thanks, Alex.

2 2

[Dovecot] Possible erroneous "aborted login attempts"
by Jerrale G 31 Aug '10

31 Aug '10

Aug 29 22:51:27 server1 dovecot: imap-login: Aborted login (no auth attempts): rip=(obfuscated), lip=173.50.101.12, TLS Aug 29 22:51:27 server1 dovecot: imap-login: Login: user=........... before most every successful login, the same second of time, dovecot has the above message. This is not a huge problem but our firewall is looking for aborted login attempts, for imap/pop3 (relevant to dovecot) dos attempts and, if many people start having problems of their packets being dropped, we will have to stop looking for the statement or lower security slightly, more attempts over a period of time before filtering. However, thanks to your idle feature, there is less of these messages; so, I don't think we will have a problem. The only client that doesn't have a problem is our php webmail but we don't look in dovecot's log for failed attempts from here; as, it is from the same ip, to dovecot from php, constantly without reference to the user. This has been happening since 1.2.11 with us. We don't use any imap relay but was looking into imapproxy for cacheing speed and preventing advanced ddos attempts, those from users with access. Jerrale G. errale G. SC Senior Admin

2 1

[Dovecot] dovecot lda issue
by Joy 31 Aug '10

31 Aug '10

I am using dovecot lda and everything is working fine for me except the sieve rule which is built for rejecting mail from a specific sender. message is being rejected but from address is not correct in rejection mail which looks like the following :- <<>@mail.xyz.net> sieve rule :- require ["fileinto","envelope","reject","vacation","imap4flags","relational","comparator-i;ascii-numeric","regex","body","date"]; if address :contains "From" "xyz(a)gmail.com" { reject text: Please do not send mail to me piyush .... . ; } Please let me know where is the issue with sieve rule or with lda setting :- lda has been configured with the following parameter :- postmaster_address = postmaster(a)xyz.net

2 1

[Dovecot] dovecot - mac firewall problem
by Patrick Fay 31 Aug '10

31 Aug '10

> >>>> Hi, >>>> I am running dovecot 1.2.11 on mac osx 1.5.8. Everything works >>>> perfectly with the application-level firewall off, but enabling the >>>> application firewall prevents dovecot connections. I have tried >>>> explicitly authorizing dovecot in the firewall, but it does not work. I >>>> have searched everywhere I can think of to look, and haven't found a >>>> solution, but have seen a couple other reports of what seems to be the >>>> same problem. The firewall logs the activity with what looks like a >>>> corrupt process name: a typical appfirewall.log entry looks like: >>>> >>>> Aug 26 20:43:45 hostname Firewall[55]: Deny ^L connecting from >>>> XX.XX.XX.XX:37310 uid = 0 proto=6 >>>> Aug 26 20:43:53 hostname Firewall[55]: Deny ^H?^U???^Z connecting from >>>> XX.XX.XX.XX:37310 uid = 0 proto=6 >>>> Aug 26 20:44:09 hostname Firewall[55]: Deny ^L connecting from >>>> XX.XX.XX.XX:37310 uid = 0 proto=6 >>>> Aug 26 20:44:34 hostname Firewall[55]: Deny ^L connecting from >>>> XX.XX.XX.XX:37312 uid = 0 proto=6 >>>> Aug 26 20:44:45: --- last message repeated 6 times --- >>>> >>>> where "hostname" is my server name and the XX's are my client's IP >>>> address. For all of the other services I've used, the process name >>>> (e.g. dovecot) should appear after "Deny" when blocking traffic, instead >>>> of the funny characters. Any advice on how I could resolve this issue >>>> would be greatly appreciated. Thanks! >>> >>> >> >> I was hoping to use the application firewall because this machine gets used both as a server as well as a client machine for more general use. I haven't been able to find any specific documentation for it, but I have found that the firewall works fine with postfix and several file services I use (enabling/disabling works as expected, process names get logged as expected, etc). Thanks! >> >> Patrick > > So, you're running Dovecot and Postfix on a laptop? WTF? > > -- > Stan Hi, Sorry for the confusion--no laptop involved. Postfix, dovecot, etc, all running on intel-based desktop mac (a mac pro). Patrick

5 5

[Dovecot] mixing 1.2 and 2.0?
by Cor Bosman 31 Aug '10

31 Aug '10

Hi all, anyone know if it's safe to mix a 1.2 environment with 2.0 servers? Im planning on adding some 2.0 servers for test purposes, but now im wondering if thats going to mess up index files or other files for users selecting the test server and then switching back to our normal servers. Regards, Cor

2 1

Re: [Dovecot] dovecot - mac firewall problem
by Patrick Fay 31 Aug '10

31 Aug '10

Hi, Thanks everyone for your help and input. I think perhaps my attempt to be brief and focused in posing my question has led to some confusion about my needs and configuration. I do not have the luxury of being able to segregate the server (postfix and dovecot) from client (GUI, etc) on different machines. I am, as Charles' surmised, using this to aggregate disparate mail streams into a single location, and using dovecot to serve it. Since I must run client-ish applications on this machine, the application firewall is desirable. The anticipated load on this system from the server side is quite light, so the throughput overhead incurred by the application firewall is negligible. The problem appears to be that--for some reason--dovecot identifies itself incorrectly to the application firewall, resulting in the garbage shown in the logs and failure to permit this connection (the firewall is configured to explicitly permit connections for dovecot). I note that other daemons (e.g. postfix, sshd, etc) do not exhibit this defect; the firewall works as expected for every other service I've tried except dovecot. The logs for the firewall appear as: Aug 26 20:43:45 hostname Firewall[55]: Deny ^L connecting from XX.XX.XX.XX:37310 uid = 0 proto=6 Aug 26 20:43:53 hostname Firewall[55]: Deny ^H?^U???^Z connecting from XX.XX.XX.XX:37310 uid = 0 proto=6 Both of these are dovecot hits--but the name is different each time. Also the ?'s aren't really "?" marks; they're trans-ascii characters with high bits set that my mailer doesn't like very much. It really looks like a misdirected pointer or something somewhere, but I'm not familiar enough with the codebase to know where to start looking, or if in fact it is something else that is misconfigured. Any thoughts on how to resolve this? Patrick

3 2

[Dovecot] getdents() with 4KB buffer - seems slow (Maildir, large inbox)
by Marinos Yannikos 31 Aug '10

31 Aug '10

Hi, I have a very large inbox (~146K mails) in Maildir format and dovecot seems to spend a lot of time rescanning the directory, especially when the server is loaded. I'm not sure whether this is triggered by Thunderbird or done regularly, but it takes longer when the server is loaded, so sometimes it seems that it is scanning continuously. Since it takes around 2000 getdents64() syscalls to scan my inbox, I'm wondering whether the buffer size is a bit small with 4KB and might benefit from being configurable: ... getdents64(8, /* 73 entries */, 4096) = 4088 getdents64(8, /* 73 entries */, 4096) = 4088 getdents64(8, /* 73 entries */, 4096) = 4088 getdents64(8, /* 73 entries */, 4096) = 4088 getdents64(8, /* 73 entries */, 4096) = 4088 ... (strace output of an imap process belonging to me) The behaviour is identical on 1.0.15 and 1.2.13 (I just upgraded). Regards, Marinos PS. I know that the main reason for slowness is my large inbox and the server being loaded (disk I/O), but still, so many syscalls just to rescan the inbox seems wasteful. # dovecot -n # 1.2.13: /etc/dovecot/dovecot.conf # OS: Linux 2.6.18-6-xen-686 i686 Debian 4.0 log_timestamp: %Y-%m-%d %H:%M:%S ssl_listen: * ssl_cert_file: XXXXX (censored) ssl_key_file: XXXXX login_dir: /var/run/dovecot/login login_executable: /usr/lib/dovecot/imap-login login_process_size: 32 login_processes_count: 8 verbose_proctitle: yes mail_privileged_group: mail mail_location: maildir:~/Maildir mbox_write_locks: fcntl dotlock mail_process_size: 2048 imap_client_workarounds: outlook-idle auth default: passdb: driver: pam userdb: driver: passwd socket: type: listen client: path: /var/spool/postfix/private/auth mode: 432 user: postfix group: postfix

2 1

[Dovecot] Mail error in log
by Ian macdonald 31 Aug '10

31 Aug '10

Hi All, I am getting the following in the mail error log deliver(julie): safe_mkstemp(/home/julie/.temp.geomail.10443.) failed: Permission denied This is correct julie does not have access to write to her home directory Julie is a system user who is chrooted so her home is owned by root, this is for when she logs in via sftp to retrieve files she can only access folders below her home. I am using Maildir and julie does have access to ~Maildir, is there any reason why deliver is attempting to use her user directory and not her Maildir? Cheers Ian

3 3

[Dovecot] istream_read like zlib, but without zlib
by Alex Baule 31 Aug '10

31 Aug '10

Hy Timo ! I Made some modification in stream_read in zlib. I remove all zlib part, because i don't need this, but i need to read a istream to change it. Well, i create a size_t called supersize, with is a substitute for stream->zs.avail_in. The trouble is, my debug file have a lot of "READ Plugin\n", and i think it's because my read becomes a loop, i think it's because i don't know to identify the EOF from the istream. you can help me to identify the EOF and set it ? Tks .!!! static ssize_t i_stream_emx_read(struct istream_private *stream) { struct emx_istream *emxstream = (struct emx_istream *)stream; const unsigned char *data; uoff_t high_offset; size_t size; int ret; fprintf(emxstream->debug,"READ Plugin\n"); fflush(emxstream->debug); high_offset = stream->istream.v_offset + (stream->pos - stream->skip); if (emxstream->eof_offset == high_offset) { i_assert(emxstream->high_pos == 0); stream->istream.eof = TRUE; return -1; } if (stream->pos < emxstream->high_pos) { /* we're here because we seeked back within the read buffer. */ ret = emxstream->high_pos - stream->pos; stream->pos = emxstream->high_pos; emxstream->high_pos = 0; return ret; } emxstream->high_pos = 0; if (stream->pos + CHUNK_SIZE > stream->buffer_size) { /* try to keep at least CHUNK_SIZE available */ if (!emxstream->marked && stream->skip > 0) { /* don't try to keep anything cached if we don't have a seek mark. */ i_stream_compress(stream); } if (stream->max_buffer_size == 0 || stream->buffer_size < stream->max_buffer_size) i_stream_grow_buffer(stream, CHUNK_SIZE); if (stream->pos == stream->buffer_size) { if (stream->skip > 0) { /* lose our buffer cache */ i_stream_compress(stream); } if (stream->pos == stream->buffer_size) return -2; /* buffer full */ } } if(emxstream->supersize == 0){ /* need to read more data. try to read a full CHUNK_SIZE */ i_stream_skip(stream->parent, emxstream->prev_size); if (i_stream_read_data(stream->parent, &data, &size, CHUNK_SIZE-1) == -1 && size == 0) { if (stream->parent->stream_errno != 0) { stream->istream.stream_errno = stream->parent->stream_errno; } else { i_assert(stream->parent->eof); stream->istream.stream_errno = EPIPE; } return -1; } emxstream->prev_size = size; if (size == 0) { /* no more input */ stream->istream.eof = TRUE; i_assert(!stream->istream.blocking); return 0; } fprintf(emxstream->debug,"READ =|%s|= Plugin\n",data); fflush(emxstream->debug); memcpy(stream->w_buffer + stream->pos, data,size); emxstream->supersize = size; } size = stream->buffer_size - stream->pos; stream->pos += size; if(stream->istream.eof == TRUE){ emxstream->eof_offset = stream->istream.v_offset + stream->pos; i_stream_skip(stream->parent, emxstream->prev_size - emxstream->supersize); emxstream->supersize = 0; emxstream->prev_size = 0; } if (size == 0) { /* read more input */ return i_stream_emx_read(stream); } return size; }

1 0