- dovecot - dovecot.org

dovecot reload 2.4.1 => core dump
by Genes Lists 16 Jul '25

16 Jul '25

"systemctl reload docecot" crashed. systemctl start and restart work fine before and after the reload crash. Running on archlinux (using the arch dovecot 2.4.1 package). Probably about 15-20 connected clients at time of crash . Is this a known issue I wonder? thanks gene All I have at this point is systemd reported trace of the crash: Process 1381 (dovecot) of user 0 dumped core. Stack trace of thread 1381: #0 0x00007e78fad7ed1d n/a (libc.so.6 + 0x16ed1d) #1 0x00007e78fac758d8 n/a (libc.so.6 + 0x658d8) #2 0x00007e78fac99e2e n/a (libc.so.6 + 0x89e2e) #3 0x00007e78fac705d6 snprintf (libc.so.6 + 0x605d6) #4 0x00007e78fad271e1 n/a (libc.so.6 + 0x1171e1) #5 0x00007e78fad27943 __syslog_chk (libc.so.6 + 0x117943) #6 0x00007e78faf38ce5 n/a (libdovecot.so.0 + 0x138ce5) #7 0x00007e78faf36d4d n/a (libdovecot.so.0 + 0x136d4d) #8 0x00007e78faf3701e i_syslog_error_handler (libdovecot.so.0 + 0x13701e) #9 0x000058ca208bd7bf n/a (/usr/bin/dovecot + 0x87bf) #10 0x00007e78faf355e6 i_warning (libdovecot.so.0 + 0x1355e6) #11 0x000058ca208c79fa n/a (/usr/bin/dovecot + 0x129fa) #12 0x00007e78faf5b698 n/a (libdovecot.so.0 + 0x15b698) #13 0x00007e78faf536e4 io_loop_call_io (libdovecot.so.0 + 0x1536e4) #14 0x00007e78faf54c8b io_loop_handler_run_internal (libdovecot.so.0 + 0x154c8b) #15 0x00007e78faf54d47 io_loop_handler_run (libdovecot.so.0 + 0x154d47) #16 0x00007e78faf54f49 io_loop_run (libdovecot.so.0 + 0x154f49) #17 0x00007e78fae9fc00 master_service_run (libdovecot.so.0 + 0x9fc00) #18 0x000058ca208bd017 main (/usr/bin/dovecot + 0x8017) #19 0x00007e78fac376b5 n/a (libc.so.6 + 0x276b5) #20 0x00007e78fac37769 __libc_start_main (libc.so.6 + 0x27769) #21 0x000058ca208bd605 _start (/usr/bin/dovecot + 0x8605) ELF object binary architecture: AMD x86-64 -- Gene "systemctl reload docecot" crashed. systemctl start and restart work fine before and after the reload crash. Running on archlinux (using the arch dovecot 2.4.1 package). Probably about 15-20 connected clients at time of crash . Is this a known issue I wonder? thanks gene All I have at this point is systemd reported trace of the crash: Process 1381 (dovecot) of user 0 dumped core. Stack trace of thread 1381: #0 0x00007e78fad7ed1d n/a (libc.so.6 + 0x16ed1d) #1 0x00007e78fac758d8 n/a (libc.so.6 + 0x658d8) #2 0x00007e78fac99e2e n/a (libc.so.6 + 0x89e2e) #3 0x00007e78fac705d6 snprintf (libc.so.6 + 0x605d6) #4 0x00007e78fad271e1 n/a (libc.so.6 + 0x1171e1) #5 0x00007e78fad27943 __syslog_chk (libc.so.6 + 0x117943) #6 0x00007e78faf38ce5 n/a (libdovecot.so.0 + 0x138ce5) #7 0x00007e78faf36d4d n/a (libdovecot.so.0 + 0x136d4d) #8 0x00007e78faf3701e i_syslog_error_handler (libdovecot.so.0 + 0x13701e) #9 0x000058ca208bd7bf n/a (/usr/bin/dovecot + 0x87bf) #10 0x00007e78faf355e6 i_warning (libdovecot.so.0 + 0x1355e6) #11 0x000058ca208c79fa n/a (/usr/bin/dovecot + 0x129fa) #12 0x00007e78faf5b698 n/a (libdovecot.so.0 + 0x15b698) #13 0x00007e78faf536e4 io_loop_call_io (libdovecot.so.0 + 0x1536e4) #14 0x00007e78faf54c8b io_loop_handler_run_internal (libdovecot.so.0 + 0x154c8b) #15 0x00007e78faf54d47 io_loop_handler_run (libdovecot.so.0 + 0x154d47) #16 0x00007e78faf54f49 io_loop_run (libdovecot.so.0 + 0x154f49) #17 0x00007e78fae9fc00 master_service_run (libdovecot.so.0 + 0x9fc00) #18 0x000058ca208bd017 main (/usr/bin/dovecot + 0x8017) #19 0x00007e78fac376b5 n/a (libc.so.6 + 0x276b5) #20 0x00007e78fac37769 __libc_start_main (libc.so.6 + 0x27769) #21 0x000058ca208bd605 _start (/usr/bin/dovecot + 0x8605) ELF object binary architecture: AMD x86-64 -- Gene

2 3

Any planned 2.4.2 release date
by Andreas Haerter 14 Jul '25

14 Jul '25

Hi, I am currently planning the migration from Dovecot 2.3 to 2.4 and was wondering if version 2.4.2 is just around the corner (and it therefore might make sense to wait a few more days). Is there any planned schedule or does it still make sense to migrate to 2.4.1 within the next ~14 days? -- Regards, Andreas Haerter foundata GmbH Steinhäuserstr. 20 76135 Karlsruhe Sitz der Gesellschaft: Karlsruhe Registergericht: Amtsgericht Mannheim, HRB 714807 Geschäftsführung: Andreas Haerter USt-IdNr.: DE284122682

10 20

Access to deleted emails
by Kamil Jońca 14 Jul '25

14 Jul '25

Some time ago I mistakenly delete a lot of emails (mdbox storage) I was able to recover them with command like: --8<---------------cut here---------------start------------->8--- doveadm import mdbox_deleted:~/Mail/0 "" ALL --8<---------------cut here---------------end--------------->8--- But now, with I get: --8<---------------cut here---------------start------------->8--- %doveadm import -u kjonca mdbox_deleted:~/Mail/0 "" ALL doveadm(kjonca): Fatal: Import namespace initialization failed: mdbox_deleted: mdbox: mailbox_root_directory_name must not be empty --8<---------------cut here---------------end--------------->8--- How can I restore mails? KJ -- http://wolnelektury.pl/wesprzyj/teraz/ Hartley's First Law: You can lead a horse to water, but if you can get him to float on his back, you've got something.

1 0

Better 2.4.x docs?
by Steve Litt 13 Jul '25

13 Jul '25

Are you planning to improve your Dovecot 2.4.x documentation? The current 2.4.x documentation has: * Non-obvious links between 2.4.x and 2.3.x * Some 2.3.x specific grammar * Almost no examples * 2.4.x material is scattered across many pages * Contains a few inaccuracies * Contains some contradictions 2.4.x is such a huge change that it should have its own, standalone documentation tree exclusively for 2.4.x. This tree should have a different color scheme than your other docs so people know at a glance exactly what they're looking at. You could have a page devoted to often used dovecot.conf content, with numerous examples, documented as if there had never been an earlier version. You could have a page devoted to conversion of 2.3.x dovecot.conf to 2.4.x, once again, with numerous examples. Another page with more exotic dovecot.conf content, with numerous examples. Another page discussing the 2.3.x features that don't appear in 2.4.x, what capabilities that takes away, and how best to cobble together those capabilities for those who don't want Dovecot Professional. Another page for brand new capabilities of 2.4.x. Thanks, SteveT Steve Litt http://444domains.com

4 5

Dovecot 2.4 migration / passwd authentication
by John 12 Jul '25

12 Jul '25

Hello, I have problem getting passwd/userdb to work on 2.4. Debian 13 Trixie, Exim4, Dovecot 2.4.0. pop3, lmtp On a 2.3 system I could use: passdb { driver=passwd-file args = /etc/dovecot/passwd } userdb { driver=static args = uid=vmail gid=vmail home=/home/vmail/%u } service lmtp { unix_listener lmtp { } } passwd looks like this: test@onedomain.tlda:{BLF-CRYPT}$2y$05$[encoded] test@domain.tld:{ARGON2ID}$argon2id$v=19$m=65536,t=3,p=1$[encoded] test@two-domain.tldb:{ARGON2ID}$argon2id$v=19$m=65536,t=3,p=1$[encoded] I want to keep the passwd file as is, no extra fields. mail_uid = vmail mail_gid = vmail mail_home = /var/vmail/%{user} mail_driver = maildir mail_path = /home/vmail/%{user}/Maildir pop3_uidl_format = %{guid} passdb passwd-file { driver=passwd-file } userdb static { driver=static } Directory /var/vmail/test(a)domain.tld exists (empty). This doesn't work. Error messages: Exim log - attempt to deliver message: ** test(a)domain.tld R=dovecot_myrouter_lmtp T=dovecot_transport_lmtp: LMTP error after RCPT TO:<test(a)domain.tld>: 550 5.1.1 <test(a)domain.tld> User doesn't exist: test(a)domain.tld Dovecot debug log: master: Info: Dovecot v2.4.1-4 (7d8c0e5759) starting up for pop3, lmtp (core dumps disabled) auth(info): Info: passwd-file: unknown user - trying the next passdb lmtp(16933): Debug: Effective uid=0, gid=0, home=/run/dovecot/empty lmtp(16933): Debug: none: root=/home/vmail/raw-mail-user/Maildir, index=, indexpvt=, control=, inbox=, alt= lmtp(test(a)domain.tld)<16933><[some_string]>: Debug: auth-master: userdb lookup(test(a)domain.tld): Started userdb lookup lmtp(test(a)domain.tld)<16933><[some_string]>: Debug: auth-master: conn unix:/run/dovecot/auth-userdb: Connecting lmtp(test(a)domain.tld)<16933><[some_string]>: Debug: auth-master: conn unix:/run/dovecot/auth-userdb (pid=16922,uid=0): Client connected (fd=16) lmtp(test(a)domain.tld)<16933><[some_string]>: Debug: auth-master: conn unix:/run/dovecot/auth-userdb (pid=16922,uid=0): auth input: lmtp(test(a)domain.tld)<16933><[some_string]>: Debug: auth-master: userdb lookup(test(a)domain.tld): auth USER input: lmtp(test(a)domain.tld)<16933><[some_string]>: Debug: auth-master: userdb lookup(test(a)domain.tld): Userdb lookup failed lmtp(16933): Debug: User session is finished # doveadm user test(a)domain.tld field value uid vmail gid vmail home /home/vmail/test(a)domain.tld mail_path /home/vmail/test(a)domain.tld/Maildir How can this be fixed? Thank you for any help! John Sent with Proton Mail secure email.

2 3

Doveadm HTTP API authCacheFlush error
by Ricardo Machini 08 Jul '25

08 Jul '25

Hello, I'm getting a fatal error when I try to run the authCacheFlush command via the doveadm HTTP API. Example of the call: curl -u user:password -X POST http://localhost:9010/doveadm/v1 -H "Content-Type: application/json" -d '[["authCacheFlush",{"user":"username"},"tag1"]]' Fatal error: doveadm(::1): Panic: file doveadm-cmd-parse.c: line 120 (doveadm_cmd_param_str): assertion failed: (param->type == CMD_PARAM_STR) doveadm(::1): Error: Raw backtrace: #0 test_subprocess_fork[0x7fc5fd934a70] -> #1 backtrace_append[0x7fc5fd934d10] -> #2 backtrace_get[0x7fc5fd934eb0] -> #3 event_filter_register_cmp[0x7fc5fd93ea50] -> #4 i_syslog_fatal_handler[0x7fc5fd93eb80] -> #5 i_panic[0x7fc5fd86eca3] -> #6 [unw_get_proc_name() failed: -10] -> #7 http_server_connection_shut_down[0x7fc5fd8dd1b0] -> #8 io_loop_call_io[0x7fc5fd959a40] -> #9 io_loop_handler_run_internal[0x7fc5fd95ad80] -> #10 io_loop_handler_run[0x7fc5fd95af20] -> #11 io_loop_run[0x7fc5fd95b110] -> #12 master_service_run[0x7fc5fd8af640] -> #13 main[0x5593ce2ab3d0] -> #14 __libc_start_call_main[0x7fc5fd429550] -> #15 __libc_start_main[0x7fc5fd429600] -> #16 _start[0x5593ce2ab5c0] doveadm(::1): Fatal: master: service(doveadm): child 168532 killed with signal 6 (core dumped) Regards, Ricardo Hello, I'm getting a fatal error when I try to run the authCacheFlush command via the doveadm HTTP API. Example of the call: curl -u user:password -X POST [1]http://localhost:9010/doveadm/v1 -H "Content-Type: application/json" -d '[["authCacheFlush",{"user":"username"},"tag1"]]' Fatal error: doveadm(::1): Panic: file doveadm-cmd-parse.c: line 120 (doveadm_cmd_param_str): assertion failed: (param->type == CMD_PARAM_STR) doveadm(::1): Error: Raw backtrace: #0 test_subprocess_fork[0x7fc5fd934a70] -> #1 backtrace_append[0x7fc5fd934d10] -> #2 backtrace_get[0x7fc5fd934eb0] -> #3 event_filter_register_cmp[0x7fc5fd93ea50] -> #4 i_syslog_fatal_handler[0x7fc5fd93eb80] -> #5 i_panic[0x7fc5fd86eca3] -> #6 [unw_get_proc_name() failed: -10] -> #7 http_server_connection_shut_down[0x7fc5fd8dd1b0] -> #8 io_loop_call_io[0x7fc5fd959a40] -> #9 io_loop_handler_run_internal[0x7fc5fd95ad80] -> #10 io_loop_handler_run[0x7fc5fd95af20] -> #11 io_loop_run[0x7fc5fd95b110] -> #12 master_service_run[0x7fc5fd8af640] -> #13 main[0x5593ce2ab3d0] -> #14 __libc_start_call_main[0x7fc5fd429550] -> #15 __libc_start_main[0x7fc5fd429600] -> #16 _start[0x5593ce2ab5c0] doveadm(::1): Fatal: master: service(doveadm): child 168532 killed with signal 6 (core dumped) Regards, Ricardo References Visible links 1. http://localhost:9010/doveadm/v1

1 0

Re: How to require client SSL certificate, except for local connections
by Nikolaos Chatzikonstantinou 08 Jul '25

08 Jul '25

Hello all, Is there a solution to this message: <https://dovecot.org/list/dovecot/2017-October/109576.html>? Regards, Nikolaos Chatzikonstantinou

1 0

Access issue with public/shared mailboxes after migrating to v2.4.x
by Juergen Edner 08 Jul '25

08 Jul '25

Hi all, I'm currently trying to get my public/shared mailbox access working again after migration to 2.4.1. I thought I would have done all required changes, but Dovecot 2.4.1 seems to act slightly different than 2.3.x. I've tried to understand what is wrong, read the documentation multiple times and searched the internet for a solution, but unfortunately without final success. I hope you can shed some light on how to get the original behavior back. Let me explain my original setup, which worked for years without any issues. I've setup the following shared folders as an adminstrator and set the access right on a directly level, e.g. /home/imapshared imapshared:users 0770 /home/imapshared/.Parents imapshared:parents 0770 /home/imapshared/.Kids imapshared:kids 0770 This was my working Dovecot 2.3 configuration: namespace { type = shared separator = / prefix = '#Shared/' location = maildir:/home/imapshared:INDEXPVT=~/.imapmail/shared subscriptions = no } User, who were allowed to access the folder, got e.g. the 'parents' or 'kids' group assigned as secondary group, the primary one is 'users'. Based on the documentation I've modified the Dovecot 2.4 configuration as follows: namespace shared { type = shared mail_driver = maildir separator = / prefix = '#Shared/' mail_path = /home/imapshared mail_index_private_path = ~/.imapmail/shared subscriptions = no } This is what I already found out or tried to adjust: - I found out that the shared folder can be accessed if the shared folder system group is set to 'users' (primary group), but by default not if a system group is used which is the secondary group of a user (e.g. parents or kids) - In the documentation I found a hint, that Dovecot by default would now only accept the primary group. By setting the following parameter additional groups should also be accepted: mail_access_groups = yes Unfortunately, that doesn't restore the orinal functionality. - In the documentation I found a hint, that all folders which have been created by an administrator, should now be of type 'public' but changing that setting doesn't work either. "It's important that the namespace type is "public" regardless of whether you set the namespace prefix to "shared." or something else." This is the error which is logged, which is not true: Error: maildir: stat(/home/imapshared/.Parents/tmp) failed: Permission denied (euid=2000(father) egid=100(users) missing +x perm: /home/imapshared/.Parents, we're not in group 203(parents), dir owned by 2226:203 mode=0770) This is the user used to access the folder: uid=2000(father) gid=100(users) groups=100(users), ..., 203(parents), ... Do you have any idea or hint how to solve this issue and provide an example what needs to be modified in my configuration to get the original functionality back? Thanks Juergen -- Mail: juergen.edner(a)telejeck.de GPG Key available -- Mail: juergen(a)eisfair.org

1 0

Pigeonhole releases
by Aleksandr Mezin 07 Jul '25

07 Jul '25

Hi. I've been using https://pigeonhole.dovecot.org/releases/2.4/ to monitor when new releases are published. Now that URL suddenly returns 403. Was it an intentional change? Why? Also, https://pigeonhole.dovecot.org/download.html still points to v2.4.0 https://pigeonhole.dovecot.org/releases/2.4/dovecot-pigeonhole-2.4.0.tar.gz as the latest release. Even though a newer v2.4.1-4 https://pigeonhole.dovecot.org/releases/2.4/dovecot-pigeonhole-2.4.1-4.tar.… was published more than 2 months ago. Also, is it v2.4.1-4 or v2.4.1? The Git tag is `2.4.1` https://github.com/dovecot/pigeonhole/releases/tag/2.4.1 but the attached tarball name ends with `-2.4.1-4` (as well as the name of the tarball on https://pigeonhole.dovecot.org/releases/) What does the `-4` suffix mean? Will it be present in future versions?

2 1

Re: USERDB/LDAP fails if attribute doesn't exist
by ricardomachini＠gmail.com 06 Jul '25

06 Jul '25

Same problem here. I would like to set some default value when attribute is not found on LDAP. Luis, did you find out how to solve it? Regards, Ricardo Same problem = here. I would like = to set some default value when attribute is not found on = LDAP. Luis, did you find out how to solve = it? Regards, Ricardo

1 0