October 2006 - dovecot

[Dovecot] Deliver: command time exceed limit
by Dominique Feyer 12 Nov '06

12 Nov '06

We use Dovecot on a cluster (10'000 domains, 30'000 account) without any problem with pop/imap. We use Dovecot 1.0 RC10 Now we use virtual transport from postfix for mail delivery. We need the Dovecot LDA for sieve support. We test this solution on some domains with success. But if we change the configuration of all our domains, we have a lots of bounce with this error in postfix log: postfix/pipe[24573]: 4F6D7CF11: to=<info(a)XXXX.com>, relay=dovecot, delay=1200, status=bounced (Command time limit exceeded: "/usr/local/dovecot/libexec/dovecot/deliver") If I watch the deliver activity with ps/awk/watch, I see that if deliver duration time exceed some seconds, the deliver process stay in memory but do nothing. Here a strace output of this case: Process 349 attached - interrupt to quit gettimeofday({1161965973, 768478}, {0, 0}) = 0 poll( Strace output nothing before the SIGKILL from postfix (command time exceed 1200 second), the end of the strace: [{fd=5, events=POLLIN|POLLPRI|POLLERR|POLLHUP|POLLNVAL}, {fd=9, events=POLLIN|POLLPRI|POLLERR|POLLHUP|POLLNVAL}], 2, 2147483647) = -1 EINTR (Interrupted system call) +++ killed by SIGKILL +++ I'm not sure is this problem has a relation with our deliver problem. But in the deliver log /var/log/dovecot/deliver, I have a lots of message: deliver(info(a)labomex.com): 2006.10.27 18:29:28 Error: file_dotlock_replace(/var/mail/labomex.com/mails/info/.dovecot.lda-dupes) failed: No such file or directory deliver(info(a)labomex.com): 2006.10.27 18:29:28 Error: rename(/var/mail/labomex.com/mails/info/.dovecot.lda-dupes.lock, /var/mail/labomex.com/mails/info/.dovecot.lda-dupes) failed: No such file or directory Our dovecot.conf (lda parts): protocol lda { postmaster_address = postmaster(a)clm.net4all.ch #hostname = mail_plugins = cmusieve mail_plugin_dir = /usr/local/dovecot/lib/dovecot/lda sendmail_path = /usr/sbin/sendmail auth_socket_path = /var/run/dovecot/auth-master log_timestamp = %Y.%m.%d %H:%M:%S%t log_path = /var/log/dovecot/deliver info_log_path = /var/log/dovecot/deliver.info } We use on UID/GID per domain so every account of a domain use the same UID/GID. The mail storage is on a Debian Sarge NFS server with 2.4.27-3-686-smp. Indexes are stored localy on each POP/IMAP/LDA server. I don't understand why we have this problem. If somebody can help us, thanks a lots. Dominique Feyer

2 7

[Dovecot] Corrupted index cache file dovecot.index.cache: invalid record size
by bofh list 07 Nov '06

07 Nov '06

Hi, Our dovecot setup consists of two hosts running dovecot-1.0.beta9 with Maildir/indices stored on NFS(noac,actimeo=0 used). I am seeing these messages at times - but no real problems on the client side. Is this something to worry about? dovecot: Oct 17 10:33:31 Error: IMAP(user): Corrupted index cache file mailstore/user/Maildir/.mail.incoming/dovecot.index.cache: invalid record size Thanks!

2 6

[Dovecot] RC10 errors
by Eric Rostetter 07 Nov '06

07 Nov '06

I just installed a limited test of RC10 last night. So far, all is great except for errors being logged for one user (out of about 150 users total). This one user has not yet complained or anything, but the log file has repeated entries of the following format for him (host and username changed to protect the guilty/innocent): Oct 24 06:13:02 mail dovecot: IMAP(username): file mbox-sync-rewrite.c: line 405 (mbox_sync_read_and_move): assertion failed: (need_space == (uoff_t)-mails[idx].space) Oct 24 06:13:02 mail1 dovecot: child 18235 (imap) killed with signal 6 I find it most interesting that only 1 out of 150+ users is getting this error... And again, no complaints from the user. Setup is: wu-imap mbox files in /var/spool/mail (exported via NFS, but local to the dovecot server). Mail in ~/mail/ (NFS mount, foreign to devecot server). Indexes in /var/dovecot/indexes (local to dovecot server with no NFS at all). -- Eric Rostetter The Department of Physics The University of Texas at Austin Go Longhorns!

1 1

[Dovecot] 1.0rc10 proxy memory usage problem
by Laurent Papier 07 Nov '06

07 Nov '06

Hi, I have a rather large dovecot setup (4 proxy only servers and 21 IMAP/POP3 servers) in production for 3 days now. The memory usage on my proxy servers only grows over time. It looks like a memory leak somewhere in dovecot. I have a RRD graph available on demand. Any advise on the configuration ? Timo, I have strace a process before the manual restart. If it can help just ask for. My configuration : Linux kernel 2.6. I use MySQL as authentication back-end. # dovecot --version 1.0.rc10 # dovecot --build-options Build options: ioloop=epoll notify=dnotify ipv6 openssl SQL drivers: mysql Passdb: checkpassword pam passwd passwd-file shadow sql Userdb: checkpassword passwd prefetch passwd-file sql static ----- config ----- protocols = pop3 imap disable_plaintext_auth = no ssl_disable = yes login_process_size = 128 login_process_per_connection = no login_processes_count = 16 login_max_connections = 512 login_greeting = ready auth_cache_size = 32768 auth_cache_ttl = 1800 auth_worker_max_count = 30 auth default { mechanisms = plain # Userdb settings are not used with proxy but there need to be something. userdb static { args = static uid=5000 gid=5000 home=/dev/null } passdb sql { args = /etc/dovecot-mysql.conf } user = root count = 1 } -- Laurent Papier

2 5

[Dovecot] suspected bug in dovecot-ldap setup?
by Pedro Venda 07 Nov '06

07 Nov '06

Hi everyone, [first post: long and boring. beware!] I've been working on a virtual multidomain MX server with pop3 access. Accounting for users/domains is done via OpenLDAP with the Jamm[1] schema and dovecot is 1.0rc10 (dovecot-1.0-0_27.rc10.el4.at fetched as RPM from atrpms.net for CentOS 4.4). I intended to have authenticated binds but there was a problem with the user_filter directive being ignored: user_filter = (&(objectClass=JammMailAccount)(mail=%n@%d)(accountActive=TRUE) (delete=FALSE)) dovecot: Oct 26 16:20:58 Info: auth(default): client in: AUTH 1 PLAIN service=IMAP secured lip=::ffff:192.168.0.5 rip=::ffff:192.168.0.69 resp=<hidden> dovecot: Oct 26 16:20:58 Info: auth(default): ldap(dave@info.test,::ffff:192.168.0.69): bind search: base=o=hosting,dc=example,dc=com scope=subtree filter= **(&(objectClass=posixAccount(uid=dave(a)info.test))** dovecot: Oct 26 16:20:58 Info: auth(default): ldap(dave@info.test,::ffff:192.168.0.69): unknown user dovecot: Oct 26 16:20:59 Info: auth(default): client out: FAIL 1 user=dave(a)info.test dovecot: Oct 26 16:20:59 Info: imap-login: Disconnected: user=<dave(a)info.test>, method=PLAIN, rip=::ffff:192.168.0.69, lip=::ffff:192.168.0.5, TLS (nevermind the data, it's test only) As you can see from the info file, the search filter used was being (&(objectClass=posixAccount(uid=dave(a)info.test)) and not (&(objectClass=JammMailAccount)(mail=dave(a)info.test)(accountActive=TRUE) (delete=FALSE)). I sorted this out by trying to enable auth_bind_userdn: auth_bind_userdn = mail=%n@%d,jvd=%d,o=hosting,dc=example,dc=com Now, the login worked well and the debug info is as follows: dovecot: Oct 27 12:01:48 Info: auth(default): client in: AUTH 1 PLAIN service=IMAP lip=::ffff:192.168.0.5 rip=::ffff:192.168.0.69 resp=AGRhdmVAaW5mb y50ZXN0AFRlbXAuMTIz dovecot: Oct 27 12:01:48 Info: auth(default): client out: OK 1 user=dave(a)info.test dovecot: Oct 27 12:01:48 Info: auth(default): master in: REQUEST 1 10634 1 dovecot: Oct 27 12:01:48 Info: auth(default): ldap(dave@info.test,::ffff:192.168.0.69): base=o=hosting,dc=example,dc=com scope=subtree filter=(&(objectClass=JammMailAccount)(mail=dave(a)info.test) (accountActive=TRUE)(delete=FALSE)) fields=mailbox dovecot: Oct 27 12:01:48 Info: auth(default): master out: USER 1 dave(a)info.test mail=info.test/dave/ uid=5000 gid=5000 (nevermind the debug data, passwords, etc. it's test only) In this case, the search filter is ok: (&(objectClass=JammMailAccount)(mail=dave(a)info.test)(accountActive=TRUE) (delete=FALSE)) So aparently, the auth_bind_userdn directive that supposedly adds a performance gain by sparing one bind request, seems to interfere with the user_filter directive. I suspect this might be a bug, so here's my report. Is this a new issue? Thanks for surviving this far through this post, Best regards, Pedro Venda. [1]: [Java Mail Manager]: web applications to manage virtual email account information stored in an LDAP directory. (http://jamm.sourceforge.net) -- Pedro João Lopes Venda email: pjvenda at pjvenda org http://www.pjvenda.org

2 5

[Dovecot] trash plugin not working as expected
by Robert Schetterer 04 Nov '06

04 Nov '06

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi @ll ,i use dovecot 1.rc7 setting in dovecot.conf protocol imap { mail_plugins = quota imap_quota } protocol pop3 { mail_plugins = quota } plugin { quota = maildir trash = /etc/dovecot/dovecot-trash.conf } dovecot-trash.conf file 1 Sent results in Disconnected: Too many invalid IMAP commands when i try to sent mails from a mailbox which is nearly full with inbox and Sent mails folder I expected the oldest mail in Sent to be deleted before copy to Sent after deliver. as described here http://wiki.dovecot.org/Plugins/Trash the client was squirrelmail 1.48 the setup works very nice without having tras installed sombody with an idea, or did i miss something in the conf or missunderstood the wiki? - -- Mit freundlichen Gruessen Best Regards Robert Schetterer robert_at_schetterer_dot_org Munich / Bavaria / Germany https://www.schetterer.org https://www.schetterer.com/public-gpg-robert-schetterer.key -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (MingW32) iD8DBQFFL9a6NxddAhXBw7QRAsusAKCeeGoK5ePoc7ccO6UHbsextBbJSwCfd1ry aBhnaGZSVizfwunNU0iZYZ4= =jtrL -----END PGP SIGNATURE----- -- Diese Nachricht wurde auf Viren und andere gefährliche Inhalte untersucht und ist - aktuelle Virenscanner vorausgesetzt - sauber.

2 1

[Dovecot] Authentication in outlook
by Paul Aguirre 03 Nov '06

03 Nov '06

Hi all, I have instaled dovecot 1.0.rc10, and my configuration works fine if I get the mail with Evolution mail client, bu if I try to use outlook to get mails the authentication fail and I can't pass through the logon window. The maillog appear this message: dovecot: pop3-login: Disconnected: rip=192.168.0.22, lip=192.168.0.1 My dovecot configuration is: protocols = pop3 ssl_disable = yes mail_extra_groups = mail protocol imap { } protocol pop3 { pop3_uidl_format = %v.%u } protocol lda { postmaster_address = postmaster(a)example.com } auth_username_translation = "@." auth default { mechanisms = plain passdb pam { } userdb passwd { } user = root } dict { } plugin { } any idea to solve. thanks Paul

6 7

[Dovecot] any idea why the mail downloading is slow?
by LeVA 03 Nov '06

03 Nov '06

Hi! I'm using rc10, but had this problem with previous releases too. Not sure if this is a dovecot problem, but I can not think of anything else. The mail downloading with pop3 is ~very slow, approx. 20 Kbyte / sec. no matter how many mails I've got in my mailbox, or how big are the mails. My ISP provides me a dedicated 100mbit line, and everything else works as fast as expected eg. http, ftp, scp... The OS is OpenBSD 3.9-stable and the system is not under heavy load at all. I've tried to turn off TLS but there was no change. Is the pop3 protocol this slow by design or there is something wrong with my setup? Thanks! Daniel -- LeVA

2 4

[Dovecot] CRAM-MD5 auth broken with postgresql passdb?
by Chris Laif 03 Nov '06

03 Nov '06

I'm currently experiencing a problem which has already been described by Jonathan in http://www.dovecot.org/list/dovecot/2006-August/015501.html If I use CRAM-MD5 auth with passwd-file as a backend (Thunderbird client) everything works fine (MD5-hashes have been shortened for readability): dovecot: auth(default): password(bob(a)foo.bar,10.0.0.123): Credentials: f12c884ba3cc82.. dovecot: imap-login: Login: user=<bob(a)foo.bar>, method=CRAM-MD5, rip=10.0.0.123, lip=10.0.0.234 My passwd file contains a line like this: bob@foo.bar:{HMAC-MD5}f12c884ba3cc82.. If I change to postgresql-based passdb, PLAIN auth works (!) and CRAM-MD5 does not work: dovecot: auth(default): sql(bob(a)foo.bar,10.0.0.123): query: SELECT 'bob(a)foo.bar' AS user, '{HMAC-MD5}f12c884ba3cc82..' AS password dovecot: auth(default): password(bob(a)foo.bar,10.0.0.123): Credentials: dovecot: auth(default): cram-md5(bob(a)foo.bar,10.0.0.123): password mismatch Please note the ultra-simple SELECT-statement (I've tried other combinations as well - no success). sql.conf setting is "default_pass_scheme = HMAC-MD5". Dovecot version 1.0.rc10. In the postgresql-based example the "Credentials:" variable is empty. I wonder if this is a bug in dovecot. Please let me know if you need any further information to resolve this issue. Thanks, Chris (As a side not I still do not understand how CRAM-MD5 auth is able to work without a plain text password, but that's another story. If it works with an HMAC-MD5 hash in a passwd-file backend it should work with the postgresql-db backend as well.)

2 3

[Dovecot] Check plugin loading for LDA and dovecot-sieve
by Will Prater 03 Nov '06

03 Nov '06

Hello, I found this thread: http://www.dovecot.org/list/dovecot/2006- September/015955.html and have successfully compiled and installed cmusieve plugin. How does one test to make sure its being loaded? It's not parsing my sieve files in the users homes and I'd like to first check if the plugin is being loaded, since I am getting no errors about it. Thanks

2 3