- dovecot - dovecot.org

SMTPUTF8 - what's the status?
by Kamil Jońca 11 Sep '25

11 Sep '25

Does dovecot LMTP can process SMTPUTF8 emails? I guess not, but I want to confirm. KJ -- http://wolnelektury.pl/wesprzyj/teraz/ <kira> is a surgical war where you go give the foreign troops nose jobs?

2 1

Re: SMTPUTF8 - what's the status?
by Michael Slusarz 11 Sep '25

11 Sep '25

> On 09/11/2025 10:53 AM MDT Kamil Jońca via dovecot <dovecot(a)dovecot.org> wrote: > > Does dovecot LMTP can process SMTPUTF8 emails? > I guess not, but I want to confirm. https://doc.dovecot.org/2.4.1/core/summaries/settings.html#mail_utf8_extens… Requires --enable-experimental-mail-utf8 when compiling. michael

1 0

Sieve Scripts via SQL Dicts in Dovecot 2.4 not working?
by Niklas Meyer 11 Sep '25

11 Sep '25

Hello everyone, I am currently working on porting the Dovecot 2.4 configuration from the latest 2.3 version. During this process, I encountered issues with the Sieve filters, particularly those related to SQL dict mappings. The previous (2.3) configuration looked like this and worked: |map { pattern = priv/sieve/name/$script_nametable = sieve_before username_field = username value_field = idfields { script_name = $script_name} } map { pattern = priv/sieve/data/$idtable = sieve_before username_field = username value_field = script_data fields { id= $id} } | This resulted in the following SQL queries on the MySQL server: |2025-08-1511:01:29.237511SELECTid FROMsieve_after WHEREscript_name ='active'ANDusername ='frittenjonas(a)derlinkman.de'2025-08-1511:01:29.236416SELECTid FROMsieve_before WHEREscript_name ='active'ANDusername ='frittenjonas(a)derlinkman.de'| As you can see, there are SQL mappings for |sieve_before| and |sieve_after|, alongside a local file loaded prior to the SQL settings, as they are user-based. In Dovecot 2.3, everything worked and the scripts were executed. In 2.4, however, they no longer run. I adapted the configuration for dict SQL mapping for Sieve filters according to the documentation: |dict_mappriv/sieve/name/$script_name{ sql_table= sieve_before username_field = username value_field id { } # The script name field in the table to querykey_fieldscript_name { value= $script_name} } dict_map priv/sieve/data/$id{ sql_table= sieve_before username_field = username value_field script_data { } key_fieldid { value= $id} } | This also results in the dict proxy querying the database with the same pattern as before: |2025-08-1511:11:47.175184SELECTid FROMsieve_after WHEREscript_name ='active'ANDusername ='frittenjonas(a)derlinkman.de'2025-08-1511:11:47.172952SELECTid FROMsieve_before WHEREscript_name ='active'ANDusername ='frittenjonas(a)derlinkman.de'2025-08-1511:11:47.155786SELECTid FROMsieve_after WHEREscript_name ='active'ANDusername ='frittenjonas(a)derlinkman.de'2025-08-1511:11:47.136030SELECTid FROMsieve_before WHEREscript_name ='active'ANDusername ='frittenjonas(a)derlinkman.de'| However, unlike in 2.3, it now queries the DB twice. In the Sieve debug logs, I see the following: https://pastebin.com/PTKkLsDQ Notably: |Aug 1512:22:2443664c142e0a dovecot: lmtp(frittenjonas(a)derlinkman.de)<2407><mPFNN2AKn2hnCQAAhwrv7A>: Debug: Localdelivery failed: Temporarily unable toaccessnecessary Sieve scripts | This line concerns me. Here are the current Sieve script declarations: |sieve_script before{ type= beforedriver = file path= /var/vmail/sieve/global_sieve_before.sieve } sieve_script before2 { type= beforedriver = dict name= active dict proxy { name= sieve_before sieve_script_bin_path = /var/vmail/sieve_before_bindir } } sieve_script after{ type= afterdriver = file path= /var/vmail/sieve/global_sieve_after.sieve } sieve_script after2 { type= afterdriver = dict name= active dict proxy { name= sieve_after sieve_script_bin_path = /var/vmail/sieve_after_bindir } } # Personal scripts sieve_script personal { type= personal driver = file path= ~/sieve active_path = ~/.dovecot.sieve } | The dict proxy configuration, pointing to the dict_map SQL query above, looks like this: |dict sieve_after { driver = sql!include/etc/dovecot/sql/dovecot-dict-sql-sieve_after.conf } dict sieve_before { driver = sql!include/etc/dovecot/sql/dovecot-dict-sql-sieve_before.conf } | Thanks for your guidance. Kind regards Niklas Meyer Maintainer of the mailcow project Hello everyone, I am currently working on porting the Dovecot 2.4 configuration from the latest 2.3 version. During this process, I encountered issues with the Sieve filters, particularly those related to SQL dict mappings. The previous (2.3) configuration looked like this and worked: map { pattern = priv/sieve/name/$script_name table = sieve_before username_field = username value_field = id fields { script_name = $script_name } } map { pattern = priv/sieve/data/$id table = sieve_before username_field = username value_field = script_data fields { id = $id } } This resulted in the following SQL queries on the MySQL server: 2025-08-15 11:01:29.237511 SELECT id FROM sieve_after WHERE script_name = 'active' AND username = '[1]frittenjonas(a)derlinkman.de' 2025-08-15 11:01:29.236416 SELECT id FROM sieve_before WHERE script_name = 'active' AND username = '[2]frittenjonas(a)derlinkman.de' As you can see, there are SQL mappings for sieve_before and sieve_after, alongside a local file loaded prior to the SQL settings, as they are user-based. In Dovecot 2.3, everything worked and the scripts were executed. In 2.4, however, they no longer run. I adapted the configuration for dict SQL mapping for Sieve filters according to the documentation: dict_map priv/sieve/name/$script_name { sql_table = sieve_before username_field = username value_field id { } # The script name field in the table to query key_field script_name { value = $script_name } } dict_map priv/sieve/data/$id { sql_table = sieve_before username_field = username value_field script_data { } key_field id { value = $id } } This also results in the dict proxy querying the database with the same pattern as before: 2025-08-15 11:11:47.175184 SELECT id FROM sieve_after WHERE script_name = 'active' AND username = '[3]frittenjonas(a)derlinkman.de' 2025-08-15 11:11:47.172952 SELECT id FROM sieve_before WHERE script_name = 'active' AND username = '[4]frittenjonas(a)derlinkman.de' 2025-08-15 11:11:47.155786 SELECT id FROM sieve_after WHERE script_name = 'active' AND username = '[5]frittenjonas(a)derlinkman.de' 2025-08-15 11:11:47.136030 SELECT id FROM sieve_before WHERE script_name = 'active' AND username = '[6]frittenjonas(a)derlinkman.de' However, unlike in 2.3, it now queries the DB twice. In the Sieve debug logs, I see the following: [7]https://pastebin.com/PTKkLsDQ Notably: Aug 15 12:22:24 43664c142e0a dovecot: lmtp([8]frittenjonas(a)derlinkman.de)<2407><mPFNN2AKn2hnCQAAhwrv7A>: Debug: Local delivery failed: Temporarily unable to access necessary Sieve scripts This line concerns me. Here are the current Sieve script declarations: sieve_script before { type = before driver = file path = /var/vmail/sieve/global_sieve_before.sieve } sieve_script before2 { type = before driver = dict name = active dict proxy { name = sieve_before sieve_script_bin_path = /var/vmail/sieve_before_bindir } } sieve_script after { type = after driver = file path = /var/vmail/sieve/global_sieve_after.sieve } sieve_script after2 { type = after driver = dict name = active dict proxy { name = sieve_after sieve_script_bin_path = /var/vmail/sieve_after_bindir } } # Personal scripts sieve_script personal { type = personal driver = file path = ~/sieve active_path = ~/.dovecot.sieve } The dict proxy configuration, pointing to the dict_map SQL query above, looks like this: dict sieve_after { driver = sql !include /etc/dovecot/sql/dovecot-dict-sql-sieve_after.conf } dict sieve_before { driver = sql !include /etc/dovecot/sql/dovecot-dict-sql-sieve_before.conf } Thanks for your guidance. Kind regards Niklas Meyer Maintainer of the mailcow project References Visible links 1. mailto:frittenjonas@derlinkman.de 2. mailto:frittenjonas@derlinkman.de 3. mailto:frittenjonas@derlinkman.de 4. mailto:frittenjonas@derlinkman.de 5. mailto:frittenjonas@derlinkman.de 6. mailto:frittenjonas@derlinkman.de 7. https://pastebin.com/PTKkLsDQ 8. mailto:frittenjonas@derlinkman.de

4 8

Sample with spool file: .mbsyncrc, Isync, Information Lifecycle Management, dovecot, GMail, IMAP
by Philip Bondi 10 Sep '25

10 Sep '25

Hello to all: I wanted to offer my working .mbsyncrc to the community. I also have a spool file to share. In my opinion, 15 Gb on our Google accounts is no longer enough. Google Photos keeps prompting to enable backup. Most people are going to exceed this quota. In my opinion, this is manifestation of serious "platform decay". This is starting to affect all my immediate family members. In my opinion, one should have methodology to move GMail away of Google for Information Lifecycle Management. To achieve, I have setup Debian 12 virtual machine on QNAP running Dovecot. This setup is beyond the scope of this article. I setup Isync to sync gmail to my Debian 12 virtual machine. Then I wrote a script to move the gmail to local mbox. Isync will then purge the email from Gmail. https://www.snbforums.com/threads/sample-with-spool-file-mbsyncrc-isync-inf…

1 0

Red Hat Enterprise Linux 8 packages
by Aki Tuomi 08 Sep '25

08 Sep '25

Hi! Due to algorithm limitations, we have signed the RHEL8 packages for 2.4.1 with Dovecot 2.3 signing key. This should make it verifiable on RHEL8. Regards, Aki Tuomi Open-Xchange oy

1 0

RHEL 8 key import error
by Gioele Pannetto 08 Sep '25

08 Sep '25

Hi, I'm having issues while installing Dovecot 2.4 on a Rocky Linux 8.10 server. This server was running Dovecot 2.3 but I wanted to upgrade it to 2.4. This is my repo file: [dovecot-2.4-latest] name=Dovecot 2.4 RHEL $releasever - $basearch baseurl=https://repo.dovecot.org/ce-2.4-latest/rhel/$releasever/RPMS/$basea… gpgkey=https://repo.dovecot.org/DOVECOT-REPO-GPG-2.4 gpgcheck=1 enabled=1 When I run dnf update I get a GPG error: Problem opening package dovecot-2.4.1-4.x86_64.rpm Problem opening package dovecot-lua-2.4.1-4.x86_64.rpm Problem opening package dovecot-mysql-2.4.1-4.x86_64.rpm The downloaded packages were saved in cache until the next successful transaction. You can remove cached packages by executing 'dnf clean packages'. Error: GPG check FAILED If I try to import the key manually with rpm --import https://repo.dovecot.org/DOVECOT-REPO-GPG-2.4 I get: error: https://repo.dovecot.org/DOVECOT-REPO-GPG-2.4: key 1 import failed. I also tried it on a fresh Rocky Linux 8.10 installation and I get the same exact error, but running it on Rocky Linux 9.6 works just fine. This is my gpg --version on RHEL 8.10: gpg (GnuPG) 2.2.20 libgcrypt 1.8.5 Copyright (C) 2020 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <https://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Home: /root/.gnupg Supported algorithms: Pubkey: RSA, ELG, DSA, ECDH, ECDSA, EDDSA Cipher: IDEA, 3DES, CAST5, BLOWFISH, AES, AES192, AES256, TWOFISH, CAMELLIA128, CAMELLIA192, CAMELLIA256 Hash: SHA1, RIPEMD160, SHA256, SHA384, SHA512, SHA224 Compression: Uncompressed, ZIP, ZLIB, BZIP2 Any help is appreciated. Thanks! -- Gioele Pannetto

6 8

Is it possible to use literal %2F inside a dovecot.http.client request url?
by hmmmz1000＠gmail.com 07 Sep '25

07 Sep '25

Hello, I'm using dovecot (2.3) lua with dovecot.http.client but I need the request url to have an url encoded slash (/) as %2F (or %2f) but if I enter for example https://test.com/%2F/test it looks like it changes it to a slash (https://test.com///test) in the http debug log (also the request fails as it's not %2F). I've also tried with %252F to encode the % but that doesn't change at all (keeps being %252F) & also tried with %%2F and \%2F but that gets the errors "Expecting hex digit after '%', but found '%'" for %%2F and "invalid escape sequence" for \%2F. Does anyone have an idea how to solve this or if it's even possible?

2 1

Help: Sharing IMAP folder from a mdbox mailbox to a maildir mailbox
by Zhang Huangbin 05 Sep '25

05 Sep '25

Dear all, I'm configuring Dovecot 2.4.1 with per-user mailbox format support, but cannot get user shared mailbox working between different mailbox formats. - Account mdbox(a)domain.com is a mdbox format mailbox, access with webmail is ok. - Account maildir(a)domain.com is a maildir format mailbox, access with webmail is ok. - mdbox@ shares its Sent folder to maildir@ with Roundcube webmail, grants all privileges (`akxeilprwts`). The result is maildir@ cannot see the shared folder. - No error reported in Dovecot log file. - mysql logs the queries performed to get user shared mailbox, i ran the query manually and it returns correct result like this: ``` MariaDB [vmail]> SELECT dummy,from_user FROM share_folder WHERE to_user = 'maildir(a)domain.com' \G *************************** 1. row *************************** dummy: 1 from_user: mdbox(a)domain.com ``` dovecot.conf snippet attached below. I think the "mail_*" parameters inside `namespace shared {}` block are incorrect to handle mixed mailbox formats because it hard-codes the mail_driver and mail_path, tried all possible parameters i can imagine, but failed. Thanks for helping. :) # ------------------------ mail_plugins { acl = yes ... } protocol imap { mail_plugins { imap_acl = yes ... } ... } mail_driver = %{userdb:mail_driver | default("maildir") | lower} mail_path = %{userdb:mail_path | default("~/Maildir")} sql_driver = mysql mysql 127.0.0.1 { port = ... dbname = ... user = ... password = ... } userdb sql { # ... omit other lines ... query = SELECT \ LOWER('%{user | lower}') AS master_user, \ LOWER(CONCAT(mailbox.storagebasedirectory, '/', mailbox.storagenode, '/', mailbox.maildir)) AS home, \ mailbox.mailboxformat AS mail_driver, \ CONCAT("~/", mailbox.mailboxfolder) AS mail_path, \ CONCAT('*:bytes=', mailbox.quota*1048576) AS quota_rule \ FROM ... } namespace shared { type = shared separator = / prefix = Shared/$user/ subscriptions = yes list = children mail_driver = maildir mail_path = %{owner_home}/Maildir mail_index_private_path = ~/Maildir/shared/%{owner_user} # not share the index files between users mail_index_path = ~/Maildir/shared/%{owner_user} } acl_driver = vfile acl_globals_only = no acl_sharing_map { dict proxy { name = acl } } dict_server { dict acl { driver = sql dict_map shared/shared-boxes/user/$to/$from { sql_table = share_folder value_field dummy { } key_field from_user { value = $from } key_field to_user { value = $to } } dict_map shared/shared-boxes/anyone/$from { sql_table = anyone_shares value_field dummy { } key_field from_user { value = $from } } } ... } # ------------------------

4 11

High Available (SoHo) dovecot setup using PVE+Ubuntu (+Docker)?
by Gerben Wierda 04 Sep '25

04 Sep '25

I currently run a HA postfix/dovecot setup that is failover behind a HAproxy. The two legs are a dovecot 2 on macOS (MacPorts) and a dovecot 2 on Ubuntu (PVE+Ubuntu+Docker), making the system somewhat antifragile. HA is realised via dovecot’s own syncing (which is not longer available in dovecot 3). This works OK. (In addition: there is rspamd as an important part of the setup, e.g. for DMARC). There are multiple virtual domains. Given that I have lost APNS (Apple Push Notification) as Apple has retired support for former Mac OS X Server users, there is not longer a reason to keep the macOS version running (apart from that antifragility thing, but that was more an added bonus). I am looking at moving to a different setup where the entire mail setup may move to a full Linux foundation. I am starting to look at future options (there is still time to actually create the setup somewhere the coming year or so. If I want to be future-dovecot proof, I need to move to HA for storage as dovecot syncing is deprecated. I wonder what my best way forward its, based on a 2 physical machine, PVE+Ubuntu+Docker setup. Build a HA NFSv4 cluster? Use Ceph? ZFS? Any tips, experiences, or ideas are welcome. Robustness/security/low-maintenance/own-your-own-data are my main objectives. I do not have high volumes, this is a setup for just a few users. G I currently run a HA postfix/dovecot setup that is failover behind a HAproxy. The two legs are a dovecot 2 on macOS (MacPorts) and a dovecot 2 on Ubuntu (PVE+Ubuntu+Docker), making the system somewhat antifragile. HA is realised via dovecot's own syncing (which is not longer available in dovecot 3). This works OK. (In addition: there is rspamd as an important part of the setup, e.g. for DMARC). There are multiple virtual domains. Given that I have lost APNS (Apple Push Notification) as Apple has retired support for former Mac OS X Server users, there is not longer a reason to keep the macOS version running (apart from that antifragility thing, but that was more an added bonus). I am looking at moving to a different setup where the entire mail setup may move to a full Linux foundation. I am starting to look at future options (there is still time to actually create the setup somewhere the coming year or so. If I want to be future-dovecot proof, I need to move to HA for storage as dovecot syncing is deprecated. I wonder what my best way forward its, based on a 2 physical machine, PVE+Ubuntu+Docker setup. Build a HA NFSv4 cluster? Use Ceph? ZFS? Any tips, experiences, or ideas are welcome. Robustness/security/low-maintenance/own-your-own-data are my main objectives. I do not have high volumes, this is a setup for just a few users. G

5 8

How to override mail_driver, mail_path, mail_index_path with SQL userdb
by Zhang Huangbin 04 Sep '25

04 Sep '25

Dear all, I'm migrating from Dovecot 2.3 to 2.4 on Debian 13, only one issue remains: How can i override global mail_driver, mail_path, mail_index_path parameters with a SQL userdb? I have these lines in dovecot.conf to use maildir format mailbox by default, also userdb and passdb with mariadb backend. We store mailbox format (e.g. "maildir", "mdbox", case insensitive) in SQL column `mailboxformat`, and mailbox folder name in `mailboxfolder` (e.g. "Maildir", "mdbox", case sensitive). I tried returning `mail_driver` + `mail_path`, or `userdb_mail_driver` + `userdb_mail_path`, neither works. So the question is, what should i return in userdb sql query to override global setting? Thanks for helping. :) # ------------------------------- mail_driver = maildir mail_path = ~/Maildir mail_index_path = ~/Maildir sql_driver = mysql mysql 127.0.0.1 { port = 3306 dbname = ... user = ... password = ... } userdb sql { iterate_query = ... query = SELECT \ LOWER('%{user | lower}') AS master_user, \ LOWER(CONCAT(mailbox.storagebasedirectory, '/', mailbox.storagenode, '/', mailbox.maildir)) AS home, \ LOWER(mailbox.mailboxformat) AS mail_driver, \ CONCAT("~/", mailbox.mailboxfolder) AS mail_path, \ CONCAT(LOWER(mailbox.mailboxformat), ":~/", mailbox.mailboxfolder) AS mail, \ CONCAT('*:bytes=', mailbox.quota*1048576) AS quota_rule \ FROM mailbox,domain \ WHERE ... } passdb sql { query = SELECT ... } # -------------------------------

1 1