- dovecot - dovecot.org

enabling fts_flatcurve on debian
by Stefan G. Weichinger 03 Apr '26

03 Apr '26

Which xapian-related packages do I need? I have issues enabling the plugin ... pls advise

2 6

2.4 upgrade : INBOX does not exist
by Stefan G. Weichinger 02 Apr '26

02 Apr '26

Today I upgraded a server at a customer. I modified the config I had built over the last week, with the help of this list. So far nearly everything works, but thunderbird complains about non-existing mailboxes like INBOX. But we can access it, it's a bit strange. In the filesystem it looks like: # ls -la /home/vmail/our.tld/user\(a)our.tld/ insgesamt 7096 drwx------ 69 vmail vmail 4096 2. Apr 19:00 . drwx------ 129 vmail vmail 12288 2. Apr 18:56 .. drwx------ 5 vmail vmail 4096 8. Jun 2021 .Administrator drwx------ 5 vmail vmail 4096 2. Apr 18:40 .Administrator.Archiv drwx------ 5 vmail vmail 4096 2. Apr 18:40 '.Administrator.Domainwechsel AquaCC' drwx------ 5 vmail vmail 4096 11. Mai 2025 .Archives drwx------ 5 vmail vmail 4096 2. Apr 18:40 .Archives.2001 drwx------ 5 vmail vmail 4096 2. Apr 18:40 .Archives.2002 drwx------ 5 vmail vmail 4096 2. Apr 18:40 .Archives.2003 drwx------ 5 vmail vmail 4096 12. Aug 2024 .Archives.2004 drwx------ 5 vmail vmail 4096 2. Apr 18:40 .Archives.2005 drwx------ 5 vmail vmail 4096 12. Aug 2024 .Archives.2006 drwx------ 5 vmail vmail 4096 12. Aug 2024 .Archives.2007 drwx------ 5 vmail vmail 4096 10. Feb 18:17 .Archives.2008 drwx------ 5 vmail vmail 4096 12. Aug 2024 .Archives.2009 drwx------ 5 vmail vmail 4096 12. Aug 2024 .Archives.2010 drwx------ 5 vmail vmail 4096 10. Feb 18:10 .Archives.2011 drwx------ 5 vmail vmail 4096 12. Aug 2024 .Archives.2012 drwx------ 5 vmail vmail 4096 12. Aug 2024 .Archives.2013 drwx------ 5 vmail vmail 4096 10. Feb 18:09 .Archives.2014 drwx------ 5 vmail vmail 4096 12. Aug 2024 .Archives.2015 drwx------ 5 vmail vmail 4096 2. Apr 17:51 .Archives.2016 drwx------ 5 vmail vmail 4096 11. Feb 11:57 .Archives.2017 drwx------ 5 vmail vmail 4096 10. Jun 2025 .Archives.2018 drwx------ 5 vmail vmail 4096 11. Feb 11:57 .Archives.2019 drwx------ 5 vmail vmail 4096 10. Feb 18:23 .Archives.2020 drwx------ 5 vmail vmail 4096 1. Apr 20:18 .Archives.2021 drwx------ 5 vmail vmail 4096 30. Mär 09:37 .Archives.2022 drwx------ 5 vmail vmail 4096 1. Apr 20:18 .Archives.2023 drwx------ 5 vmail vmail 4096 1. Apr 20:18 .Archives.2024 drwx------ 5 vmail vmail 4096 2. Apr 17:48 .Archives.2025 drwx------ 5 vmail vmail 4096 2. Apr 17:59 .Archives.2026 drwx------ 2 vmail vmail 323584 2. Apr 19:00 cur -rw------- 1 vmail vmail 0 2. Aug 2021 dovecot-acl-list -rw------- 1 vmail vmail 70712 2. Apr 18:37 dovecot.index -rw------- 1 vmail vmail 6266860 2. Apr 19:01 dovecot.index.cache -rw------- 1 vmail vmail 32160 2. Apr 19:00 dovecot.index.log -rw------- 1 vmail vmail 46940 2. Apr 06:20 dovecot.index.log.2 -rw------- 1 vmail vmail 91 18. Aug 2020 dovecot-keywords -rw------- 1 vmail vmail 65 2. Apr 18:37 .dovecot.lda-dupes drwx------ 5 vmail vmail 4096 2. Apr 18:37 .dovecot.lda-dupes.locks -rw------- 1 vmail vmail 6504 2. Apr 18:49 dovecot.list.index -rw------- 1 vmail vmail 3356 2. Apr 19:02 dovecot.list.index.log -rw------- 1 vmail vmail 2904 7. Jän 13:53 dovecot.mailbox.log -rw------- 1 vmail vmail 186894 2. Apr 18:59 dovecot-uidlist -rw------- 1 vmail vmail 8 7. Jän 13:53 dovecot-uidvalidity -r--r--r-- 1 vmail vmail 0 21. Apr 2020 dovecot-uidvalidity.5e9efc52 drwx------ 5 vmail vmail 4096 2. Apr 12:02 .Drafts drwx------ 5 vmail vmail 4096 12. Jun 2025 '.Entw&APw-rfe' drwx------ 5 vmail vmail 4096 6. Mai 2020 .Gesendet drwx------ 5 vmail vmail 4096 2. Apr 18:22 .INBOX.Administrator drwx------ 5 vmail vmail 4096 2. Apr 18:21 .INBOX.Alerts [..] We have mail_home = /home/vmail/%{user | domain}/%{user} I tried mail_home = /home/vmail/%{user | domain}/%{user | username} but that breaks things completely. In 2.3 we had these namespaces: namespace { type = private separator = / prefix = #location defaults to mail_location. inbox = yes } namespace { type = public separator = / prefix = Oeffentlich/ location = maildir:/home/vmail/oeffentlich subscriptions = yes } I tried "prefix = INBOX/" ... not good. I share the config, a little bit anonymized, acls removed ... # doveconf -n # 2.4.1-4 (7d8c0e5759): /etc/dovecot/dovecot.conf # Pigeonhole version 2.4.1-4 (0a86619f) # OS: Linux 6.12.74+deb13+1-amd64 x86_64 Debian 13.4 # Hostname: xxxx dovecot_config_version = 2.4.1 auth_allow_weak_schemes = yes auth_cache_size = 4M auth_debug = yes auth_mechanisms = plain login auth_verbose = yes default_process_limit = 300 dovecot_storage_version = 2.4.0 import_environment { CORE_ERROR = %{env:CORE_ERROR} CORE_OUTOFMEM = %{env:CORE_OUTOFMEM} LISTEN_FDS = %{env:LISTEN_FDS} LISTEN_PID = %{env:LISTEN_PID} MALLOC_MMAP_THRESHOLD_ = 131072 NOTIFY_SOCKET = %{env:NOTIFY_SOCKET} PATH = %{env:PATH} TZ = :/etc/localtime } info_log_path = /var/log/dovecot.info listen = * [::] log_path = /var/log/dovecot.log mail_debug = yes mail_driver = maildir mail_gid = 5000 mail_home = /home/vmail/%{user | domain}/%{user} mail_path = ~ mail_plugins { acl = yes notify = yes quota = no } mail_uid = 5000 protocols = imap sieve lmtp sieve_extensions { fileinto = yes reject = yes envelope = yes encoded-character = yes vacation = yes subaddress = yes comparator-i;ascii-numeric = yes relational = yes regex = yes imap4flags = yes copy = yes include = yes body = yes variables = yes enotify = yes environment = yes mailbox = yes date = yes index = yes ihave = yes duplicate = yes mime = yes foreverypart = yes extracttext = yes editheader = yes mboxmetadata = yes servermetadata = yes spamtest = yes spamtestplus = yes virustest = yes } sieve_global_extensions { vnd.dovecot.debug = yes vnd.dovecot.environment = yes vnd.dovecot.pipe = yes } sieve_pipe_bin_dir = /usr/lib/dovecot/sieve sieve_plugins = sieve_imapsieve sieve_extprograms sieve_spamtest_score_max_value = 15 sieve_spamtest_status_header = X-Spamd-Result: .*? \[([[:digit:]]+\.[[:digit:]]+) / [[:digit:]]+\.[[:digit:]]+\]; sql_driver = mysql ssl = required verbose_proctitle = yes mysql /var/run/mysqld/mysqld.sock { dbname = postfix_db password = # hidden, use -P to show it user = postfix_user } passdb sql { default_password_scheme = MD5-CRYPT query = SELECT username as user, password FROM mailbox WHERE ((username = '%{user | username}(a)domain1.at') OR (username = '%{user | username}(a)domain2.at')) AND active = '1' } userdb sql { iterate_query = SELECT userid AS username, domain FROM users query = SELECT username as user FROM mailbox WHERE ((username = '%{user | username}(a)domain1.at') OR (username = '%{user | username}(a)domain2.at') ) AND active = '1' } namespace inbox { inbox = yes prefix = separator = / type = private } namespace oeffentlich { mail_path = /home/vmail/oeffentlich prefix = Oeffentlich/ separator = / subscriptions = yes type = public mailbox Oeffentlich { acl [DELETED for now] } service imap-login { process_min_avail = 8 inet_listener imap { port = 0 ssl = no listen = } inet_listener imaps { port = 993 ssl = yes listen = } } service lmtp { extra_groups = $SET:default_internal_group user = vmail unix_listener /var/spool/postfix/private/dovecot-lmtp { group = postfix mode = 0660 user = postfix } } service imap { } service auth { client_limit = 1200 user = root unix_listener /var/spool/postfix/private/auth { group = postfix mode = 0666 user = postfix } unix_listener auth-client { group = postfix mode = 0660 user = postfix } unix_listener auth-userdb { mode = 0660 } } service auth-worker { } service dict { unix_listener dict { } } ssl_server { cert_file = /etc/letsencrypt/pfsense/cloud_imap_mail_smtp_LE_cert.fullchain key_file = /etc/letsencrypt/pfsense/cloud_imap_mail_smtp_LE_cert.key } protocol lda { mail_plugins = sieve } protocol imap { imap_idle_notify_interval = 29 mins mail_max_userip_connections = 50 mail_plugins { imap_acl = yes imap_quota = no imap_sieve = yes } } protocol lmtp { mail_plugins = sieve postmaster_address = logs(a)xunil.at } service managesieve-login { inet_listener sieve { port = 4190 } } service managesieve { } protocol sieve { managesieve_notify_capability = mailto managesieve_sieve_capability = fileinto reject envelope encoded-character vacation subaddress comparator-i;ascii-numeric relational regex imap4flags copy include variables body enotify environment mailbox date ihave } sieve_script personal { active_path = /home/vmail/%{user | domain}/%{user}/sieve/scripts/.dovecot.sieve driver = file path = /home/vmail/%{user | domain}/%{user}/sieve/scripts } sieve_script before-0 { driver = file path = /usr/lib/dovecot/sieve/global-spam.sieve type = before } service anvil { unix_listener anvil { group = $SET:default_internal_group mode = 0660 } } I assume I am not far from full success, I'd just like to get rid of these errors because tomorrow dozens of users will access their mailboxes ... Thanks for any pointers!

1 1

BUG-Report: mysql: SSL support not compiled in (remove ssl_client_ca_file and ssl_client_ca_dir settings) third attempt!
by Klaus Tachtler 02 Apr '26

02 Apr '26

Hello everyone, Problem info: ============ An error occurs, while trying to establish a TLS/SSL secured connection to a database (mariadb), which has the ability to accept TLS/SSL connections. Analysis from Archlinux Gitlab Issue, which could be found under following link by - Toolybird: https://gitlab.archlinux.org/archlinux/packaging/packages/dovecot/-/issues/… > Thanks for reporting. It would appear to be some kind of (upstream?) > bug in the autofoo. The error message can be traced back to > HAVE_MYSQL_SSL being undefined when it should be defined. This is > reflected in log: https://reproducible.archlinux.org/api/v0/builds/946041/log > It took quite some debugging, but adding the following hack before the > autoreconf call appears to make it work: > checking for mysql_ssl_set in -llibmariadb... no > sed -i 's/mysql_driver="libmariadb"/mysql_driver="mariadb"/' m4/ > want_mysql.m4 > After that it now shows: > checking for mysql_ssl_set in -lmariadb... yes Steps to reproduce: ================== 1. Configure dovecot 2.4.2-4 as described below 2. Restart dovecot 2.4.2-4 3. Try send an email (echo "Test-E-Mail" | /usr/sbin/sendmail root) 4. Determine error messages inside dovecot 2.4.2-4-Log > 01 Dec 22 04:57:38 server dovecot[842]: auth: Error: > mysql(db.idmz.tachtler.net): mysql: SSL support not compiled in > (remove ssl_client_ca_file and ssl_client_ca_dir settings) > 02 Dec 22 04:57:38 server dovecot[842]: auth-worker: Error: > mysql(db.idmz.tachtler.net): mysql: SSL support not compiled in > (remove ssl_client_ca_file and ssl_client_ca_dir settings) Additional info: =============== OS version: Archlinux package version(s): dovecot 2.4.2-4 necessary config file part: > 45 sql_driver = mysql > 46 > 47 mysql db.idmz.tachtler.net { > 48 host = db.idmz.tachtler.net > 49 port = 3306 > 50 user = postfixuser > 51 password = secret > 52 dbname = postfix > 53 ssl = yes > 54 ssl_client_ca_file = /etc/dovecot/ssl/certs/CAcert.pem > 55 ssl_client_cert_file = > /etc/dovecot/ssl/certs/client.idmz.tachtler.net.pem > 56 ssl_client_key_file = > /etc/dovecot/ssl/private/client.idmz.tachtler.net.key.pem > 57 ssl_cipher_list = DHE-RSA-AES256-SHA > 58 } WITHOUT the following line, a database connections is possible and works as expected: > 53 ssl = yes > 54 ssl_client_ca_file = /etc/dovecot/ssl/certs/CAcert.pem > 55 ssl_client_cert_file = > /etc/dovecot/ssl/certs/client.idmz.tachtler.net.pem > 56 ssl_client_key_file = > /etc/dovecot/ssl/private/client.idmz.tachtler.net.key.pem > 57 ssl_cipher_list = DHE-RSA-AES256-SHA Thanky you in advance for your help! Klaus Tachtler. -- --------------------------------------- e-Mail : klaus(a)tachtler.net Homepage: https://www.tachtler.net DokuWiki: https://dokuwiki.tachtler.net ---------------------------------------

3 17

Error: sieve: fail2ban: line 23: read(<lmtp DATA local>) failed: Seeking past EOF by 2364 bytes
by Graham Cobb 01 Apr '26

01 Apr '26

Hi, dovecot is crashing for me while doing sieve processing for messages received via lmtp. This is reproducible, and seems to hit every message which runs a particular sieve script. Any thoughts or help would be appreciated! The error is: Error: sieve: fail2ban: line 23: read(<lmtp DATA local>) failed: Seeking past EOF by 2364 bytes Note: the number of bytes is always one of: 2169, 2272, 2259, 2364. I have many of these. I cannot see how the "Seeking past EOF" error can be caused by a bug in my sieve script. But please let me know if it can! Background: I have (and have had for many years) a complex set of sieve scripts for controlling my incoming email. They have worked fine for all that time. However, recently (ish), processing which uses a particular script is crashing dovecot. The actual log extract is below. The sieve script line 23, mentioned in the error report, is a large "if anyof (" statement containing many clauses like: body :raw :regex "country:[[:space:]]+KR", I can provide the full sieve script if required. This is what I see in the dovecot log (note: I reformatted the backtrace to make it easier to read. I also changed all my server names): Mar 29 03:03:52 SERVER dovecot-HOST: lmtp(4008073): Connect from local Mar 29 03:03:52 SERVER dovecot-HOST: lmtp(cobb)<4008073><PiAQHoiIyGmJKD0Ay7SJiA>: Error: sieve: fail2ban: line 23: read(<lmtp DATA local>) failed: Seeking past EOF by 2364 bytes Mar 29 03:03:52 SERVER dovecot-HOST: lmtp(cobb)<4008073><PiAQHoiIyGmJKD0Ay7SJiA>: Error: sieve: Execution of script 'auto:personal/.dovecot' was aborted due to temporary failure (user logfile /home/cobb/.dovecot.sieve.log may reveal additional details) Mar 29 03:03:52 SERVER dovecot-HOST: lmtp(cobb)<4008073><PiAQHoiIyGmJKD0Ay7SJiA>: Panic: file istream-concat.c: line 27 (i_stream_concat_close): assertion failed: (cstream->cur_input == cstream->input[cstream->cur_idx]) Mar 29 03:03:52 SERVER dovecot-HOST: lmtp(cobb)<4008073><PiAQHoiIyGmJKD0Ay7SJiA>: Error: Raw backtrace: #1 [unw_get_proc_name() failed: -10] -> #2 [unw_get_proc_name() failed: -10] -> #3 i_panic[0x7f41540655e6] -> #4 [unw_get_proc_name() failed: -10] -> #5 io_stream_unref[0x7f415413c120] -> #6 i_stream_unref[0x7f4154145c30] -> #7 index_storage_mailbox_close[0x7f4154383e70] -> #8 mailbox_close[0x7f41542fd9e0] -> #9 mailbox_free[0x7f4154302ab0] -> #10 lmtp_local_deinit[0x55dfa9c4e320] -> #11 client_state_reset[0x55dfa9c4f0d0] -> #12 smtp_server_transaction_free[0x7f4154091880] -> #13 smtp_server_connection_reset_state[0x7f4154093c90] -> #14 smtp_server_command_call_hooks[0x7f415408be60] -> #15 smtp_server_command_send_replies[0x7f4154090050] -> #16 [unw_get_proc_name() failed: -10] -> #17 [unw_get_proc_name() failed: -10] -> #18 io_loop_call_io[0x7f4154151710] -> #19 io_loop_handler_run_internal[0x7f41541534b0] -> #20 io_loop_handler_run[0x7f4154153640] -> #21 io_loop_run[0x7f4154153820] -> #22 master_service_run[0x7f41540a6330] -> #23 main[0x55dfa9c4a260] -> #24 __libc_start_call_main[0x7f4153e32c30] -> #25 __libc_start_main_alias_2[0x7f4153e32ce0] -> #26 _start[0x55dfa9c4a470] Mar 29 03:03:54 SERVER dovecot-HOST: lmtp(cobb)<4008073><PiAQHoiIyGmJKD0Ay7SJiA>: Fatal: master: service(lmtp): child 4008073 killed with signal 6 (core dumped) Mar 29 03:04:05 SERVER postfix/lmtp[4008072]: 005083DC75D: to=<MYADDRESS@MYDOMAIN>, orig_to=<ANOTHER@ANOTHERDOMAIN>, relay=SERVERNAME[private/dovecot-lmtp], delay=1740676, delays=1740674/0.25/0.04/1.5, dsn=4.4.2, status=deferred (lost connection with SERVERNAME[private/dovecot-lmtp] while sending end of data -- message may be sent more than once) The only message in .dovecot.sieve.log is: fail2ban: line 23: error: failed to read input message: internal error occurred: refer to server log for more information. [2026-04-01 18:54:13]. Thanks in advance for any thoughts! Graham

2 1

Broken table in dovecot-pw manual
by Célestin Matte 01 Apr '26

01 Apr '26

Hello, I just noticed while reading the dovecot-pw man on Arch Linux that the table in the -r option is broken (missing newline between row): | Scheme | Minimum | Maximum | Default | | ------ | ------- | ------- | ------- | | BLF-CRYPT | 4 | 31 | 5 | | SHA256-CRYPT | 1000 | 999999999 | 5000 | | SHA512-CRYPT | 1000 | 999999999 | 5000 | Regards, -- Célestin Matte

2 1

dovecot.list.index.log owned by root: why?
by Stefan G. Weichinger 01 Apr '26

01 Apr '26

I use these sql-queries to let all mailboxes be accessed by the uid=1001: passdb sql { default_password_scheme = MD5-CRYPT passdb_sql_query = SELECT username as user, password, \ '/home/vmail/%{user | domain }/%{user}' as userdb_home, 'maildir:/home/vmail/%{user | domain }/%{user}' as userdb_mail,\ 1001 as userdb_uid, 1001 as userdb_gid FROM mailbox WHERE username = '%{user}' AND active = '1' } userdb prefetch { } ( I assume this following query isn't used then anymore) userdb sql { userdb_sql_query = SELECT username as user, password, \ '/home/vmail/%{user | domain }/%{user}' as userdb_home, 'maildir:/home/vmail/%{user | domain }/%{user}' as userdb_mail,\ 1001 as userdb_uid, 1001 as userdb_gid \ FROM mailbox WHERE username = '%{user}' AND active = '1' userdb_sql_iterate_query = SELECT userid AS username, domain FROM users } id 1001 uid=1001(vmail) gid=1001(vmail) groups=1001(vmail),118(dovecot) I get errors like: Apr 01 11:43:05 imap(logs(a)xunil.at)<1564491><N8oZ6WJO/uwgAQRwHwsMRX9ul+gRMpfY>: Error: open(/home/vmail/xunil.at/logs(a)xunil.at/dovecot.list.index.log) failed: Permission denied (euid=1001(vmail) egid=1001(vmail) missing +r perm: /home/vmail/xunil.at/logs(a)xunil.at/dovecot.list.index.log) ls -l /home/vmail/xunil.at/logs(a)xunil.at/ total 944 drwx------ 2 vmail vmail 622592 Apr 1 11:43 cur -rw------- 1 vmail vmail 0 Jul 30 2021 dovecot-acl-list -rw------- 1 vmail vmail 3304 Apr 1 11:35 dovecot.index -rw------- 1 vmail vmail 202444 Apr 1 11:43 dovecot.index.cache -rw------- 1 vmail vmail 2908 Apr 1 11:43 dovecot.index.log -rw------- 1 vmail vmail 32892 Apr 1 11:35 dovecot.index.log.2 -rw------- 1 vmail vmail 91 Feb 9 2024 dovecot-keywords -rw------- 1 root root 5744 Apr 1 10:30 dovecot.list.index -rw------- 1 root root 480 Apr 1 10:52 dovecot.list.index.log [..] Why do these 2 files get owned by root now ... ? All files below "/home/vmail" should be owned by vmail:vmail I ran chown, and chgrp, sure .. but the errors return. How to fix this? Thanks in advance!

2 5

cannot get IOS or MacOSX to connect to dovecot:995 since upgraded Mac or iPhone
by Bill Scheding 01 Apr '26

01 Apr '26

Versions: [wls(a)www.wls.org]$ uname -a Linux www.wls.org 3.10.0-1160.119.1.el7.x86_64 #1 SMP Tue Jun 4 14:43:51 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux [wls(a)ww.wls.org]$ dovecot --version 2.2.36 (1f10bfa63) My Certs: Testing cert dovecot.key unable to load certificate 139668060415888:error:0906D06C:PEM routines:PEM_read_bio:no start line:pem_lib.c:707:Expecting: TRUSTED CERTIFICATE lrwxrwxrwx 1 root root 21 Mar 25 15:44 dovecot.key -> dovecot.key.old_sep_2 Testing chain dovecot.pem SHA1 Fingerprint=4A:29:34:1D:D5:9C:67:59:FA:31:7E:FF:F4:53:F8:DB:25:73:15:07 subject= /OU=IMAP server/CN=imap.example.com/emailAddress=postmaster@example.com issuer= /OU=IMAP server/CN=imap.example.com/emailAddress=postmaster@example.com serial=C8AA4C688A3F2638 b5bd7939 notBefore=Sep 2 20:17:33 2018 GMT notAfter=Sep 2 20:17:33 2019 GMT lrwxrwxrwx 1 root root 21 Mar 25 15:44 dovecot.pem -> dovecot.crt.old_sep_2 New Certs: Testing key ca.key.3.13.4 RSA key ok lrwxrwxrwx 1 root root 31 Mar 13 14:14 ca.key.3.13.4 -> ../../tls/private/ca.key.3.13.4 Testing chain ca.crt.3.13.4 SHA1 Fingerprint=79:69:88:2A:E0:BF:58:A8:21:E3:88:12:8D:45:20:4F:FD:68:33:63 subject= /C=XX/ST= /L= /O= /OU=IMAP server/CN=imap.example.com/emailAddress=postmaster@example.com issuer= /C=XX/ST= /L= /O= /OU=IMAP server/CN=imap.example.com/emailAddress=postmaster@example.com serial=B912DD19FEDC441C 402b4e2a notBefore=Mar 13 20:02:58 2026 GMT notAfter=Mar 13 20:02:58 2027 GMT lrwxrwxrwx 1 root root 29 Mar 13 14:13 ca.crt.3.13.4 -> ../../tls/certs/ca.crt.3.13.4 They install OK, dovecot starts; but iPhone and mac cannot connect to pop3 Cannot get my iPhone or Mac to accept new certs from lets encrypt, et. al. This is a new problem, started recently. My existing cert expired in 2016 William

2 1

Variable expansion syntax: Filter "default" throws error (unwanted)
by Johannes Nohl 01 Apr '26

01 Apr '26

I am using dovecot 2.4.2 (almost) with new config language and I have a working configuration. However, the check for disabled users throws permanently errors in the log file, whenever a user is missing the ldap attribute for disabled. This is - in my view - due to a misfunction of Filter "default". passdb ldap { driver = ldap ldap_bind = yes ldap_filter = (&(|(objectClass=posixAccount)(objectClass=mailAccount))(uid=%{user})) fields { user = %{ldap:uid} nologin = %{ldap:shadowInactive | default(null)} reason = The account for this email address is disabled. } } This results in the following log line: 2026-03-31T00:46:31.102690+02:00 mailserver dovecot: auth(test,ipv4,sasl:plain)<YE81n0VOVuofEeSD>: Error: ldap: auth_passdb_post settings: Failed to parse configuration: Failed to expand passdb_fields/nologin setting variables: ldap: No such attribute 'shadowinactive' The right behaviour would be no logging on login success. Because the Filter "default" is intended to suppress any errors ("When value is missing or empty, you can use the default filter to provide value. Missing variables will cause errors and must be negated with default. This does not apply to all providers, some providers return empty when value is missing."). https://doc.dovecot.org/2.4.2/core/settings/variables.html#variable-expansi… As you may have seen I had to borrow null-value from sql db config also to make nologin ignoring a non present attribute. This is undocumented obviously. Instead, a non-working tip is provided ("Changed: 2.4.0 Extra fields can now also be set to empty string, "). However, the following does not work: "nologin = {ldap:shadowinactive | default("")}". https://doc.dovecot.org/2.4.2/core/settings/variables.html Also, the documentation is missing/hiding the part which explains that "default()" can be used to avoid error logs. PR regarding the documentation source is sent already. See "Fails if the attribute is not present, unless the ` <missing part>" https://doc.dovecot.org/2.4.0/core/config/auth/databases/ldap.html#ldap-spe… Is there another way to verify if account is disabled? -- Johannes Nohl postmaster i-t-cloud -- Johannes Nohl postmaster(a)i-t-cloud.de I am using dovecot 2.4.2 (almost) with new config language and I have a working configuration. However, the check for disabled users throws permanently errors in the log file, whenever a user is missing the ldap attribute for disabled. This is - in my view - due to a misfunction of Filter "default". passdb ldap { driver = ldap ldap_bind = yes ldap_filter = (&(|(objectClass=posixAccount)(objectClass=mailAccount))(uid=%{user})) fields { user = %{ldap:uid} nologin = %{ldap:shadowInactive | default(null)} reason = The account for this email address is disabled. } } This results in the following log line: *2026-03-31T00:46:31.102690+02:00 mailserver dovecot: auth(test,ipv4,sasl:plain)<YE81n0VOVuofEeSD>: Error: ldap: auth_passdb_post settings: Failed to parse configuration: Failed to expand passdb_fields/nologin setting variables: ldap: No such attribute 'shadowinactive' The right behaviour would be no logging on login success. Because the Filter "default" is intended to suppress any errors ("When value is missing or empty, you can use the default filter to provide value. Missing variables will cause errors and must be negated with default. This does not apply to all providers, some providers return empty when value is missing."). https://doc.dovecot.org/2.4.2/core/settings/variables.html#variable-expansi… As you may have seen I had to borrow null-value from sql db config also to make nologin ignoring a non present attribute. This is undocumented obviously. Instead, a non-working tip is provided ("Changed: 2.4.0 Extra fields can now also be set to empty string, "). However, the following does not work: "nologin = {ldap:shadowinactive | default("")}". https://doc.dovecot.org/2.4.2/core/settings/variables.html Also, the documentation is missing/hiding the part which explains that "default()" can be used to avoid error logs. PR regarding the documentation source is sent already. See "Fails if the attribute is not present, unless the ` <missing part>" https://doc.dovecot.org/2.4.0/core/config/auth/databases/ldap.html#ldap-spe… Is there another way to verify if account is disabled? -- Johannes Nohl postmaster i-t-cloud -- Johannes Nohl postmaster(a)i-t-cloud.de

2 1

[BUG] 2.4.3 SIGSEGV in fts_solr
by Nick O'Connor 01 Apr '26

01 Apr '26

Dovecot 2.4.3 crashes with SIGSEGV in indexer-worker and lmtp during FTS Solr indexing. The crash is a use-after-free triggered when the HTTP client does a DNS lookup for the Solr host. Environment: - Dovecot 2.4.3 (Alpine 3.23 / musl) - Linux x86_64 - Filesystem: maildir on ZFS - Solr 10.0.0 - fts_solr plugin with DNS-based Solr URL dovecot -n: # 2.4.3 (e278b1e09b): /etc/dovecot/dovecot.conf # Pigeonhole version 2.4.3 (0bf9ee48) # OS: Linux 6.12.74+deb13+1-amd64 x86_64 dovecot_config_version = 2.4.0 auth_mechanisms = oauthbearer xoauth2 dovecot_storage_version = 2.4.0 fts_autoindex = yes fts_search_add_missing = yes mail_always_cache_fields = date.save mail_driver = maildir mail_plugins { fts = yes fts_solr = yes } protocols = imap lmtp sieve ssl = required fts solr { url = http://solr.example.com/solr/dovecot } (LDAP userdb/passdb, pigeonhole sieve, imapsieve for spam learning, and oauth2 passdb omitted) How to reproduce: The crash occurs in the indexer-worker when processing multiple user mailboxes sequentially for FTS indexing, using a DNS hostname for the Solr URL. Steps: 1. Configure fts_solr with a DNS-based URL (not an IP address) 2. Have the indexer-worker process mailboxes for more than one user (e.g. fts_autoindex = yes with multiple users, or manual doveadm index -A) 3. The first user session completes successfully. On the second user session, the indexer-worker crashes with SIGSEGV during the DNS lookup for the Solr host. Reproduces reliably on musl (Alpine) due to aggressive memory reuse. Likely also affects glibc but may not crash immediately due to delayed reuse of freed memory. Root cause: The solr_http_client singleton outlives the user session that created it. Its event chain holds a ref on the mail_storage_service_user event, keeping it alive after the user is freed. The SETTINGS_EVENT_INSTANCE on that event still points to the already-freed settings_instance. On the next user session, dns_lookup() walks the stale event chain and dereferences the dangling pointer in settings_instance_get(). Stack trace (from core dump, debug build with musl-dbg): #0 strlen(s=0x415641d189485741) -- SIGSEGV, freed/reused memory #1 p_strdup() strfuncs.c:56 #2 event_add_str(key="protocol", value=...) lib-event.c:1099 #3 settings_instance_get() settings.c:2521 #4 settings_get_real() settings.c:2706 #5 settings_get_full() settings.c:2915 #6 settings_get(dns_client_setting_parser_info) settings.c:2976 #7 dns_client_init() dns-lookup.c:465 #8 dns_client_init_handle_failure() dns-lookup.c:317 #9 dns_lookup("solr.example.com") dns-lookup.c:337 #10 http_client_host_shared_lookup() http-client-host.c:166 #11 http_client_host_shared_refresh() http-client-host.c:195 #12-#19 http_client queue/host/request submit chain #20 http_client_request_continue_payload() http-client-request.c:1170 #21 http_client_request_send_payload() http-client-request.c:1261 #22 solr_connection_post_more() solr-connection.c:269 Proposed fix: Destroy solr_http_client in solr_connection_deinit() so it is recreated with a fresh event chain on the next user session: --- a/src/plugins/fts-solr/solr-connection.c +++ b/src/plugins/fts-solr/solr-connection.c @@ -117,6 +117,8 @@ void solr_connection_deinit(struct solr_connection **_conn) *_conn = NULL; event_unref(&conn->event); + if (solr_http_client != NULL) + http_client_deinit(&solr_http_client); i_free(conn->http_host); i_free(conn->http_base_url); I've been running for few days with the patch applied. FTS indexing across multiple user sessions completes without crashes using a DNS-based Solr URL. Please let me know if there's anything else that would be helpful. The core files have sensitive information so I've omitted them from the report. -Nick Dovecot 2.4.3 crashes with SIGSEGV in indexer-worker and lmtp during FTS Solr indexing. The crash is a use-after-free triggered when the HTTP client does a DNS lookup for the Solr host. Environment: - Dovecot 2.4.3 (Alpine 3.23 / musl) - Linux x86_64 - Filesystem: maildir on ZFS - Solr 10.0.0 - fts_solr plugin with DNS-based Solr URL dovecot -n: # 2.4.3 (e278b1e09b): /etc/dovecot/dovecot.conf # Pigeonhole version 2.4.3 (0bf9ee48) # OS: Linux 6.12.74+deb13+1-amd64 x86_64 dovecot_config_version = 2.4.0 auth_mechanisms = oauthbearer xoauth2 dovecot_storage_version = 2.4.0 fts_autoindex = yes fts_search_add_missing = yes mail_always_cache_fields = date.save mail_driver = maildir mail_plugins { fts = yes fts_solr = yes } protocols = imap lmtp sieve ssl = required fts solr { url = http://solr.example.com/solr/dovecot } (LDAP userdb/passdb, pigeonhole sieve, imapsieve for spam learning, and oauth2 passdb omitted) How to reproduce: The crash occurs in the indexer-worker when processing multiple user mailboxes sequentially for FTS indexing, using a DNS hostname for the Solr URL. Steps: 1. Configure fts_solr with a DNS-based URL (not an IP address) 2. Have the indexer-worker process mailboxes for more than one user (e.g. fts_autoindex = yes with multiple users, or manual doveadm index -A) 3. The first user session completes successfully. On the second user session, the indexer-worker crashes with SIGSEGV during the DNS lookup for the Solr host. Reproduces reliably on musl (Alpine) due to aggressive memory reuse. Likely also affects glibc but may not crash immediately due to delayed reuse of freed memory. Root cause: The solr_http_client singleton outlives the user session that created it. Its event chain holds a ref on the mail_storage_service_user event, keeping it alive after the user is freed. The SETTINGS_EVENT_INSTANCE on that event still points to the already-freed settings_instance. On the next user session, dns_lookup() walks the stale event chain and dereferences the dangling pointer in settings_instance_get(). Stack trace (from core dump, debug build with musl-dbg): #0 strlen(s=0x415641d189485741) -- SIGSEGV, freed/reused memory #1 p_strdup() strfuncs.c:56 #2 event_add_str(key="protocol", value=...) lib-event.c:1099 #3 settings_instance_get() settings.c:2521 #4 settings_get_real() settings.c:2706 #5 settings_get_full() settings.c:2915 #6 settings_get(dns_client_setting_parser_info) settings.c:2976 #7 dns_client_init() dns-lookup.c:465 #8 dns_clien t_init_handle_failure() dns-lookup.c:317 #9 dns_lookup("solr.example.com") dns-lookup.c:337 #10 http_client_host_shared_lookup() http-client-host.c:166 #11 http_client_host_shared_refresh() http-client-host.c:195 #12-#19 http_client queue/host/request submit chain #20 http_client_request_continue_payload() http-client-request.c:1170 #21 http_client_request_send_payload() http-client-request.c:1261 #22 solr_connection_post_more() solr-connection.c:269 Proposed fix: Destroy solr_http_client in solr_connection_deinit() so it is recreated with a fresh event chain on the next user session: --- a/src/plugins/fts-solr/solr-connection.c +++ b/src/plugins/fts-solr/solr-connection.c @@ -117,6 +117,8 @@ void solr_connection_deinit(struct solr_connection **_conn) *_conn = NULL; event_unref(&conn->event); + if (solr_http_client != NULL) + http_client_deinit(&solr_http_client); i_free(conn->http_host); i_free(conn->http_base_url); I've been running for few days with the patch applied. FTS indexing across multiple user sessions completes without crashes using a DNS-based Solr URL. Please let me know if there's anything else that would be helpful. The core files have sensitive information so I've omitted them from the report. -Nick

1 0

migration from 2.3 to 2.4 : postfix can't deliver via lmtp
by Stefan G. Weichinger 31 Mar '26

31 Mar '26

I am in the process of upgrading and rewriting my config. I can access IMAP etc ... so my database access is working. I have issues with LMTP: postfix can't deliver to dovecot. The postfix-confiog wasn't touched at all. the error: Mar 27 11:46:37 co dovecot: auth(stefan(a)oops.co.at): Error: userdb: client doesn't have lookup permissions for this user: userdb reply doesn't contain uid (to bypass this check, set: service auth { unix_listener /run/dovecot/auth-userdb { mode=0777 } }) What I don't understand: I configure this: service auth { unix_listener /var/spool/postfix/private/auth { group = postfix mode = 0666 user = postfix } } service auth-worker { user = postfix } service lmtp { user = vmail unix_listener /var/spool/postfix/private/dovecot-lmtp { group = postfix mode = 0660 user = postfix } } protocol lmtp { mail_plugins = sieve postmaster_address = logs(a)xunil.at } And in postfix: virtual_transport = lmtp:unix:private/dovecot-lmtp virtual_uid_maps = static:5000 virtual_gid_maps = static:5000 local_transport = virtual local_recipient_maps = $virtual_mailbox_maps smtpd_sasl_auth_enable = yes smtpd_sasl_type = dovecot smtpd_sasl_path = private/auth Where does "/run/dovecot/auth-userdb" come from? I understand that according to my config /var/spool/postfix/private/auth should be used? ls /var/spool/postfix/private/auth -l srw-rw-rw- 1 postfix postfix 0 Mar 27 11:45 /var/spool/postfix/private/auth ls /var/spool/postfix/private/dovecot-lmtp -l srw-rw---- 1 postfix postfix 0 Mar 27 11:45 /var/spool/postfix/private/dovecot-lmtp I'd appreciate any quick help here ... while googling and testing around ... thanks!

2 17