On Tue, May 6, 2025 at 10:53 AM Timo Sirainen <timo@sirainen.com> wrote:
On 4. May 2025, at 14.24, Odhiambo Washington via dovecot < dovecot@dovecot.org> wrote:
passdb sql { default_password_scheme = SHA512 query = SELECT crypt AS password FROM users,domains WHERE
users.username = '%
{user}' AND users.enabled = '1' AND users.type='local' and domains.enabled='1' and domains.domain_id = users.domain_id }
userdb sql { query = SELECT pop as home, uid, gid FROM users WHERE username = '%{user}' iterate_query = SELECT username AS user FROM users
Not a reason for your troubles, but should you be checking the enabled=1 here also? So LMTP deliveries won't succeed for disabled users/domains.
May 04 13:09:12 auth-worker(9138): Debug: mysql(/var/run/mysqld/mysqld.sock): Finished query 'SELECT crypt AS password FROM users,domains WHERE users.username = 'joh@doe.com' AND users.enabled = '1' AND users.type='local' and domains.enabled='1' and domains.domain_id = users.domain_id' in 0 msecs May 04 13:09:12 auth-worker(joh@doe.com ,127.0.0.1)<9138><0sexkUw07I1/AAAB>: request [1]: Debug: sql: Finished passdb lookup
Okay..
May 04 13:09:12 auth-worker(9138): Debug: conn unix:auth-worker (pid=9063,uid=116): auth-worker<1>: Finished: internal_failure
There's an internal failure, but no reason for it in these logs. I'd have expected an error to be logged. Are the errors logged to a different file that you're not looking at? Check with "doveadm log find" and/or "doveadm log errors" if there's a reason for the internal failure.
I was relying on the sample configs provided at https://github.com/dovecot/tools/blob/main/dovecot-2.4.0-example-config.tar.... . BTW, tar -zxvf https://github.com/dovecot/tools/blob/main/dovecot-2.4.0-example-config.tar.... (on Debian 12) does not work for me with this file. Why.
Anyway, so when I extracted the above, I started by only changing the 10-auth.conf which contains: auth_mechanisms = plain login !include auth-sql.conf.ext
Then I created my auth-sql.conf.ext shown above. What I realized then is that the dovecot/auth service doesn't seem to be running. Not sure why.
May 06 12:02:20 pop3-login: Info: Login aborted: Logged out (auth service
reported temporary failure
doveadm log errors gives no output.
Looking at https://doc.dovecot.org/2.4.1/core/admin/running.html, and comparing with a system where I have installed Dovecot 2.4.1:
root@mail:/etc/dovecot/conf.d# ps auxw|grep "dovecot" root 9739 0.0 0.0 8872 4816 ? Ss 14:47 0:00 /usr/sbin/dovecot -F root 9741 0.0 0.0 5324 3280 ? S 14:47 0:00 dovecot/anvil root 9742 0.0 0.0 5428 3256 ? S 14:47 0:00 dovecot/log root 9743 0.0 0.1 51976 10248 ? S 14:47 0:00 dovecot/config
Why are my other services not running? For example, the dovecot/auth service is not running.
On an system running 2.3.21, I get:
wash@eu:~$ ps auxw|grep "dovecot" root 181404 0.0 0.0 8240 4408 ? Ss 14:33 0:00 /usr/sbin/dovecot -F dovecot 181408 0.0 0.0 10668 6936 ? S 14:33 0:00 dovecot/managesieve-login Debian-+ 181409 0.0 0.0 12192 8880 ? S 14:33 0:00 dovecot/lmtp -L dovecot 181410 0.0 0.0 4760 1436 ? S 14:33 0:00 dovecot/anvil root 181411 0.0 0.0 5028 2960 ? S 14:33 0:00 dovecot/log Debian-+ 181412 0.0 0.0 12192 8848 ? S 14:33 0:00 dovecot/lmtp -L Debian-+ 181413 0.0 0.0 12192 9040 ? S 14:33 0:00 dovecot/lmtp -L Debian-+ 181414 0.0 0.0 12192 8908 ? S 14:33 0:00 dovecot/lmtp -L Debian-+ 181415 0.0 0.0 12192 8876 ? S 14:33 0:00 dovecot/lmtp -L root 181416 0.0 0.0 8096 5444 ? S 14:33 0:00 dovecot/config dovecot 181418 0.0 0.0 5964 3496 ? S 14:33 0:00 dovecot/stats dovecot 181419 0.0 0.0 12216 8456 ? S 14:33 0:00 dovecot/auth
Is it normal that the dovecot/auth is not listed as a running service for
2.4.1?
I think that is one of my problems, besides also not getting any output
from doveadm log errors.
I wiped dovecot and reinstalled. My current config is minimal and I am not using the sample configs referred to above, but I still have the same problem:
root@mail:/var/www/html/exim4u/vexim2-master/setup# doveconf -n
2.4.1-4+debian12 (7d8c0e5759): /etc/dovecot/dovecot.conf
Pigeonhole version 2.4.1-4+debian12 (0a86619f)
OS: Linux 6.1.0-34-amd64 x86_64 Debian 12.10
Hostname: mail.mara.cloud
dovecot_config_version = 2.4.1 dovecot_storage_version = 2.4.1 info_log_path = /var/log/dovecot.log mail_driver = maildir mail_home = /home/%{user | username}/mail mail_inbox_path = /var/mail/%{user | username} mail_path = ~/mail mailbox_list_utf8 = yes protocols = imap lmtp sieve pop3 submission sql_driver = mysql passdb pam { } userdb passwd { } mysql /var/run/mysqld/mysqld.sock { dbname = exim4u password = # hidden, use -P to show it user = exim4u } passdb sql { default_password_scheme = SHA512 query = SELECT crypt AS password FROM users,domains WHERE users.username = '%{user}' AND users.enabled = '1' AND users.type='local' and domains.enabled='1' and domains.domain_id = users.domain_id } userdb sql { iterate_query = SELECT username AS user FROM users query = SELECT pop as home, uid, gid FROM users WHERE username = '%{user}' } namespace inbox { inbox = yes separator = / } ssl_server { cert_file = /etc/ssl/certs/ssl-cert-snakeoil.pem key_file = /etc/ssl/private/ssl-cert-snakeoil.key }
-- Best regards, Odhiambo WASHINGTON, Nairobi,KE +254 7 3200 0004/+254 7 2274 3223 In an Internet failure case, the #1 suspect is a constant: DNS. "Oh, the cruft.", egrep -v '^$|^.*#' ¯\_(ツ)_/¯ :-) [How to ask smart questions: http://www.catb.org/~esr/faqs/smart-questions.html]
On Tue, May 6, 2025 at 10:53 AM Timo Sirainen <timo@sirainen.com> wrote: On 4. May 2025, at 14.24, Odhiambo Washington via dovecot <dovecot@dovecot.org> wrote: > > passdb sql { > default_password_scheme = SHA512 > query = SELECT crypt AS password FROM users,domains WHERE users.username = '% > {user}' AND users.enabled = '1' AND users.type='local' and domains.enabled='1' > and domains.domain_id = users.domain_id > } > > userdb sql { > query = SELECT pop as home, uid, gid FROM users WHERE username = '%{user}' > iterate_query = SELECT username AS user FROM users
Not a reason for your troubles, but should you be checking the
enabled=1 here also? So LMTP deliveries won't succeed for disabled
users/domains.
> May 04 13:09:12 auth-worker(9138): Debug: mysql(/var/run/mysqld/
mysqld.sock):
> Finished query 'SELECT crypt AS password FROM users,domains WHERE
> users.username = 'joh@doe.com' AND users.enabled = '1' AND
users.type='local'
> and domains.enabled='1' and domains.domain_id = users.domain_id' in
0 msecs
> May 04 13:09:12 auth-worker
(joh@doe.com,127.0.0.1)<9138><0sexkUw07I1/AAAB>:
> request [1]: Debug: sql: Finished passdb lookup
Okay..
> May 04 13:09:12 auth-worker(9138): Debug: conn unix:auth-worker
> (pid=9063,uid=116): auth-worker<1>: Finished: internal_failure
There's an internal failure, but no reason for it in these logs. I'd
have expected an error to be logged. Are the errors logged to a
different file that you're not looking at? Check with "doveadm log
find" and/or "doveadm log errors" if there's a reason for the
internal failure.I was relying on the sample configs provided at https://github.com/dovecot/ tools/blob/main/dovecot-2.4.0-example-config.tar.gz. BTW, tar -zxvf https://github.com/dovecot/tools/blob/main/dovecot-2.4.0- example-config.tar.gz (on Debian 12) does not work for me with this file. Why.
Anyway, so when I extracted the above, I started by only changing the 10- auth.conf which contains: auth_mechanisms = plain login !include auth-sql.conf.ext
Then I created my auth-sql.conf.ext shown above. What I realized then is that the dovecot/auth service doesn't seem to be running. Not sure why.
May 06 12:02:20 pop3-login: Info: Login aborted: Logged out (auth service
reported temporary failure
doveadm log errors gives no output.
Looking at https://doc.dovecot.org/2.4.1/core/admin/running.html, and comparing with a system where I have installed Dovecot 2.4.1:
root@mail:/etc/dovecot/conf.d# ps auxw|grep "dovecot" root 9739 0.0 0.0 8872 4816 ? Ss 14:47 0:00 /usr/sbin/ dovecot -F root 9741 0.0 0.0 5324 3280 ? S 14:47 0:00 dovecot/ anvil root 9742 0.0 0.0 5428 3256 ? S 14:47 0:00 dovecot/log root 9743 0.0 0.1 51976 10248 ? S 14:47 0:00 dovecot/ config
Why are my other services not running? For example, the dovecot/auth service is not running.
On an system running 2.3.21, I get:
wash@eu:~$ ps auxw|grep "dovecot" root 181404 0.0 0.0 8240 4408 ? Ss 14:33 0:00 /usr/sbin/ dovecot -F dovecot 181408 0.0 0.0 10668 6936 ? S 14:33 0:00 dovecot/ managesieve-login Debian-+ 181409 0.0 0.0 12192 8880 ? S 14:33 0:00 dovecot/lmtp -L dovecot 181410 0.0 0.0 4760 1436 ? S 14:33 0:00 dovecot/ anvil root 181411 0.0 0.0 5028 2960 ? S 14:33 0:00 dovecot/log Debian-+ 181412 0.0 0.0 12192 8848 ? S 14:33 0:00 dovecot/lmtp -L Debian-+ 181413 0.0 0.0 12192 9040 ? S 14:33 0:00 dovecot/lmtp -L Debian-+ 181414 0.0 0.0 12192 8908 ? S 14:33 0:00 dovecot/lmtp -L Debian-+ 181415 0.0 0.0 12192 8876 ? S 14:33 0:00 dovecot/lmtp -L root 181416 0.0 0.0 8096 5444 ? S 14:33 0:00 dovecot/ config dovecot 181418 0.0 0.0 5964 3496 ? S 14:33 0:00 dovecot/ stats dovecot 181419 0.0 0.0 12216 8456 ? S 14:33 0:00 dovecot/auth
Is it normal that the dovecot/auth is not listed as a running service for
2.4.1?
I think that is one of my problems, besides also not getting any output from
doveadm log errors.
I wiped dovecot and reinstalled. My current config is minimal and I am not using the sample configs referred to above, but I still have the same problem:
root@mail:/var/www/html/exim4u/vexim2-master/setup# doveconf -n
2.4.1-4+debian12 (7d8c0e5759): /etc/dovecot/dovecot.conf
Pigeonhole version 2.4.1-4+debian12 (0a86619f)
OS: Linux 6.1.0-34-amd64 x86_64 Debian 12.10
Hostname: mail.mara.cloud
dovecot_config_version = 2.4.1 dovecot_storage_version = 2.4.1 info_log_path = /var/log/dovecot.log mail_driver = maildir mail_home = /home/%{user | username}/mail mail_inbox_path = /var/mail/%{user | username} mail_path = ~/mail mailbox_list_utf8 = yes protocols = imap lmtp sieve pop3 submission sql_driver = mysql passdb pam { } userdb passwd { } mysql /var/run/mysqld/mysqld.sock { dbname = exim4u password = # hidden, use -P to show it user = exim4u } passdb sql { default_password_scheme = SHA512 query = SELECT crypt AS password FROM users,domains WHERE users.username = '% {user}' AND users.enabled = '1' AND users.type='local' and domains.enabled='1' and domains.domain_id = users.domain_id } userdb sql { iterate_query = SELECT username AS user FROM users query = SELECT pop as home, uid, gid FROM users WHERE username = '%{user}' } namespace inbox { inbox = yes separator = / } ssl_server { cert_file = /etc/ssl/certs/ssl-cert-snakeoil.pem key_file = /etc/ssl/private/ssl-cert-snakeoil.key }
-- Best regards, Odhiambo WASHINGTON, Nairobi,KE +254 7 3200 0004/+254 7 2274 3223 In an Internet failure case, the #1 suspect is a constant: DNS. "Oh, the cruft.", egrep -v '^$|^.*#' ¯\_(ツ)_/¯ :-) [How to ask smart questions: http://www.catb.org/~esr/faqs/smart- questions.html]