- dovecot - dovecot.org

[Dovecot] pop3_lock_session
by Rene Luria 22 Feb '07

22 Feb '07

Hi guys, How can I modify the timeout value for the pop3_lock_session ? I'd like it to be much bigger.

2 4

[Dovecot] crash in mail_cache_transaction_reset on rc22
by Dan Price 22 Feb '07

22 Feb '07

I've not been able to roll out rc23 yet (tonight, I hope) but I just saw a crash which I'm not sure I've seen reported before, following a server outage (that is to say, the server came back up and one of the users had a dovecot core). #0 0x0005d720 in mail_cache_transaction_reset (ctx=0xcf928) at mail-cache-transaction.c:71 No locals. #1 0x0005e8bc in mail_cache_add (ctx=0xcf928, seq=233, field=4294967295, data=0xb7600, data_size=1063) at mail-cache-transaction.c:752 file_field = 19 data_size32 = 1063 fixed_size = 4294967295 full_size = 831272 __PRETTY_FUNCTION__ = "mail_cache_add" #2 0x00057160 in index_mail_parse_header (part=0xe9, hdr=0x14, mail=0xd33e8) at index-mail-headers.c:125 data = (struct index_mail_data *) 0xd3450 cache_field_name = 0xe9 <Address 0xe9 out of bounds> field_idx = 0 count = 21 match = (uint8_t *) 0xcf928 "" __PRETTY_FUNCTION__ = "index_mail_parse_header" #3 0x00079848 in _read (stream=0xe2aa0) at istream-header-filter.c:173 ret = 0 pos = 854336 #4 0x00085818 in i_stream_read (stream=0xe2ac8) at istream.c:58 _stream = (struct _istream *) 0xcaf28 #5 0x00085ca4 in i_stream_read_data (stream=0xe2ac8, data=0xffbff5ec, size=0xffbff5e8, threshold=1) at istream.c:250 ret = 0 read_more = false #6 0x0007e8c4 in message_get_body_size (input=0xe2ac8, body=0xffbff660, has_nuls=0x0) at message-size.c:105 msg = ( const unsigned char *) 0xe2cf7 "\ndary_(ID_/CFD74Co3EWCZqYkdHlE8A)\"\r\nt>\r\nSS(a)cox.net>,\r\nil.com>,\r\n\r\n" i = 0 ---Type <return> to continue, or q <return> to quit--- size = 1 missing_cr_count = 0 __PRETTY_FUNCTION__ = "message_get_body_size" #7 0x0002e5e4 in fetch_body_header_fields (ctx=0xc6a08, mail=0xffbff660, context=0xc6c98) at imap-fetch-body.c:448 size = {physical_size = 1357, virtual_size = 0, lines = 26} old_offset = Unhandled dwarf expression opcode 0x93 -- Daniel Price - Solaris Kernel Engineering - dp(a)eng.sun.com - blogs.sun.com/dp

2 1

[Dovecot] Understanding fcntl() error messages
by Dean Manners 22 Feb '07

22 Feb '07

Hey guys, After many trials and tribulations I have now got some fairly stable NFS clients serving dovecot-lda and pop3/imap. Over a 9 hour postal/rabid stress test only saw 5 errors for the 4 servers. Wondering if anyone can shed a bit more light on what they mean exactly. I have tried to find "Resource temporarily unavailable" and "No locks available" in the src to get some understanding - however it appears these messages are from a lower level. With pop3: ------------------ Feb 22 03:36:32 mailget01 dovecot: POP3(test992(a)mailtest06.com): fcntl() failed with file /mnt/mailstore/mailtest06.com/8/test992/dovecot.index.log: Resource temporarily unavailable Feb 22 03:36:32 mailget01 dovecot: POP3(test992(a)mailtest06.com): mail_index_wait_lock_fd() failed with file /mnt/mailstore/mailtest06.com/8/test992/dovecot.index.log: Resource temporarily unavailable Feb 22 04:16:45 mailget01 dovecot: POP3(test445(a)mailtest03.com): fcntl() failed with file /mnt/mailstore/mailtest03.com/4/test445/dovecot.index.log: Resource temporarily unavailable Feb 22 04:16:45 mailget01 dovecot: POP3(test445(a)mailtest03.com): mail_index_wait_lock_fd() failed with file /mnt/mailstore/mailtest03.com/4/test445/dovecot.index.log: Resource temporarily unavailable ------------------ ------------------ Feb 22 13:07:27 mailget01 dovecot: POP3(test999(a)mailtest06.com): fcntl() failed with file /mnt/mailstore/mailtest06.com/8/test999/dovecot.index: Resource temporarily unavailable Feb 22 13:07:27 mailget01 dovecot: POP3(test999(a)mailtest06.com): fcntl() failed with file /mnt/mailstore/mailtest06.com/8/test999/dovecot.index: Resource temporarily unavailable Feb 22 13:07:27 mailget01 dovecot: POP3(test999(a)mailtest06.com): Couldn't open INBOX: Internal error occurred. Refer to server log for more information. [2007-02-22 13:07:27] Feb 22 13:07:27 mailget01 dovecot: POP3(test999(a)mailtest06.com): No INBOX for user top=0/0, retr=0/0, del=0/0, size=0 ------------------ Does this mean that pop3 was waiting for a lock, but timed out ? What are the likely affects of this on a client POPing - session times out and they need to reconnect? The "No INBOX" error is little more odd, the same user had successful login 90 minutes earlier. With deliver: ------------------ Feb 22 01:42:03 mailfilter01 deliver(test680(a)mailtest04.com): fcntl() failed with file /mnt/mailstore/mailtest04.com/10/test680/dovecot.index: No locks available Feb 22 01:42:06 mailfilter01 deliver(test265(a)mailtest03.com): fcntl() failed with file /mnt/mailstore/mailtest03.com/1/test265/dovecot.index: No locks available Feb 22 03:25:42 mailfilter02 deliver(test454(a)mailtest03.com): fcntl() failed with file /mnt/mailstore/mailtest03.com/2/test454/dovecot.index: No locks available Feb 22 03:25:45 mailfilter02 deliver(test397(a)mailtest03.com): fcntl() failed with file /mnt/mailstore/mailtest03.com/7/test397/dovecot.index: No locks available ------------------ I assume this means deliver tried to obtain a lock on the index file (probably many times) but failed. Are the likely affects that the message still gets saved to the maildir but the index doesn't get updated? Regards __________________________________________ Dean Manners

2 2

[Dovecot] critical performance problem
by Rene Luria 22 Feb '07

22 Feb '07

I have been using dovecot 1.0alpha4 + deliver/sieve in production for a few accounts (400'000). After testing 1.0.rc22 I decided to upgrade one of the imap servers... And... what a mess ! The server just gets unusable. I don't know why but with new release, it just overloads the system with I/O. It turns that with 1.0.rc22 I get 15Mo/s read on disk average, without any reason. What did really change ? How come a system (with a few thousand accounts) works perfectly and then with 1.0.rc22 gets unusable with a load approaching 300 ? If you have any ideas...

5 18

[Dovecot] 1.0.rc23 tomorrow
by Timo Sirainen 22 Feb '07

22 Feb '07

http://dovecot.org/nightly/dovecot-latest.tar.gz Would be nice to get some testing before the actual release. I've now finally gone through all the "TODO" mails in this list and my INBOX and fixed pretty much every problem I know of. There are still some mbox bugs, but I can't really get them fixed unless someone figures out how to reproduce them (http://wiki.dovecot.org/MboxProblems) I think documentation is the last thing that should be done before v1.0. I'm happy enough with the code as it is now, unless some new bugs again appear. The notable changes since rc22: * deliver doesn't ever exit with Dovecot's internal exit codes anymore. All its internal exit codes are changed to EX_TEMPFAIL. + PAM: blocking=yes in args uses an alternative way to do PAM checks. Try it if you're having problems with PAM. + userdb passwd: blocking=yes in args makes the userdb lookups be done in auth worker processes. Set it if you're doing remote NSS lookups (eg. nss_ldap problems are fixed by this). + If PAM child process hasn't responded in two minutes, send KILL signal to it (only with blocking=no) - IMAP: APPEND ate all CPU while waiting for more data from the client (broken in rc22) - mbox: When saving a message to an empty mbox file it got one UID which immediately was incremented. - mbox: Fixed some wrong "uid-last unexpectedly lost" errors. - auth cache: In some situations we crashed if passdb had extra_fields. - auth cache: Special extra_fields weren't saved to auth cache. For example allow_nets restrictions were ignored for cached entries. - A lot of initial login processes could cause auth socket errors in log file at startup, if dovecot-auth started slowly. Now the login processes are started only after dovecot-auth has finished initializing itself. - imap/pop3 proxy: Don't crash if the remote server disconnects before we're logged in. - deliver: Don't bother trying to save the mail twice into the default mailbox (eg. if it's over quota). - mmap_disable=yes + non-Linux was really slow with large dovecot.index.cache files - MySQL couldn't be used as a masterdb - Trash plugin was more or less broken - imap/pop3 couldn't load plugins if they chrooted - imap/pop3-login process could crash in some conditions - checkpassword-reply crashed if USER/HOME wasn't set

6 19

[Dovecot] --log-error and the mysterious "not starting" problem
by Marshal Newrock 22 Feb '07

22 Feb '07

I've finally upgraded to 1.0.rc21. If I run "dovecot --log-error", then dovecot fails to start. The following appears in the log: Feb 5 10:54:00 taurus1 dovecot: Dovecot v1.0.rc21 starting up Feb 5 10:54:00 taurus1 dovecot: This is Dovecot's error log Without --log-error, it will start up fine from command-line. Dovecot still fails to start, as near as I can tell, in the specific condition that it is being started from a non-interactive shell and there is a mail_plugin set in "protocol imap". mail plugins set for pop or lda do not cause this problem. The failure to start appears in only two instances that I can see: on boot, or by a monitoring script if dovecot crashes, or, since I haven't seen dovecot die, when I manually stop it to see if the monitoring program will restart it. With the above problem, if the init script has --log-error when I try to make the monitoring program restart dovecot, then nothing still gets logged. -- Marshal Newrock, Ideal Solution LLC http://www.idealso.com

2 8

[Dovecot] False "new mail" warning from Thunderbird, using RC23
by Stephen Warren 21 Feb '07

21 Feb '07

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I am continually getting false "New Mail" warning from Thunderbird. I'm running Thunderbird 1.5.0.9 on Windows, and it's talking to two different Dovecot 1.0.RC23 servers (one home server, one work server). This was happening in older Dovecot versions too, so it's not new in RC23. Very often, Thunderbird will pop up the "new mail" notification window, and put a "new mail" icon in the Windows icon bar. However, when I view the main Thunderbird window, there will be no folders with new mail. Simply changing the currently selected folder in Thunderbird will clear the tray icon. Sometimes the notification window will say something small like "7 new messages", but right now, it's showing huge numbers like "587 new messages", which could be the total message count in one of my folders... Sometimes restarting Thunderbird will stop this happening for a short while. Is this a Thunderbird bug? Am I the only one seeing this? Does it have to do with having two separate IMAP servers configured? Perhaps it's because Thunderbird is configured to cache 5 connections to the server (although I think that's the default?) Should I delete all my accounts in Thunderbird, delete the (Thunderbird) cache files etc. on disk, and re-create it all? I do notice that whenever Thunderbird starts, it displays a whole load of folders that don't exist (e.g. got renamed/deleted) and/or multiple copies (e.g. Sent, Sent-1, Sent-2), which disappear very quickly. Is this related? -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFF2+eJhk3bo0lNTrURAh9uAKCWYrcNByF+wE2e1FPvL6x8N48/uACgtwxR 07fYmnaG9P/JCBOkKscCZIA= =XWXG -----END PGP SIGNATURE-----

2 1

[Dovecot] auth_user setting?
by Joseba Torre 21 Feb '07

21 Feb '07

Hi all, I've installed rc23 in a new testing machine, and i've had some permission problems related to dovecot-master.conf. As I have my master passwords in that file, I don't want it to be world readable. dovecot -n shows that auth_user is nobody, so setting the permissions to nobody.whatever 640 solves the problem. Anyway, I've tried to change the auth_user setting with no effect. Is it configurable? Aaaaaaaaaagur. -- Joseba Torre. CIDIR Bizkaia.

1 1

[Dovecot] Wrong configure.in entry in CVS branch_1_0
by Luca Corti 21 Feb '07

21 Feb '07

This line in configure.in src/plugins/mail-log/Makefile refers to a non existing direcory in the branch_1_0 tree, which breaks build. ciao Luca

3 2

[Dovecot] Possible Bug in quota-fs.c
by Martin Konopka 20 Feb '07

20 Feb '07

Hi, I'm using filesystem quotas with dovecot 1.0rc22 on my Linux 2.4.27 box. The quota limit gets correctly reported in kilobytes to the client but the amount of used space is reported in megabytes. In quota-fs.c I noticed that only the currently used space is divided by 1024: Line 241: *value_r = dqblk.dqb_curblocks / 1024; *limit_r = dqblk.dqb_bsoftlimit; After removing /1024 everything works fine. regards, Martin

3 4