June 2023 - dovecot - dovecot.org

Many IDLE imap processes but very few connections moved to imap-hibernate
by pierre.alletru＠gmail.com 17 Jul '23

17 Jul '23

Hi Dovecot community! I have a strange issue with imap-hibernate. Or more likely it behaves as it's supposed to and it's just that I don't really understand it. I have enabled it, and it works, but for some reason hibernation happens only to a tiny amount of connections, whereas I have a lot of IDLE imap processes. And I can't find any relevant errors in the logs. ## config ## mail_debug = yes verbose_proctitle = yes $ doveconf imap_hibernate_timeout imap_hibernate_timeout = 5 secs $ doveconf -a | grep -A 21 'service imap-hibernate' service imap-hibernate { chroot = client_limit = 0 drop_priv_before_exec = no executable = imap-hibernate extra_groups = group = idle_kill = 0 privileged_group = process_limit = 0 process_min_avail = 0 protocol = imap service_count = 0 type = unix_listener imap-hibernate { group = $default_internal_group mode = 0660 user = } user = $default_internal_user vsz_limit = 18446744073709551615 B } ## imap processes ## $ ps aux | grep "dovecot/imap" | wc -l 2214 ps aux | grep "dovecot/imap" | grep IDLE | wc -l 1258 $ ps aux | grep "imap-hibernate" syslog 1571453 0.0 0.0 4748 3156 ? S 15:44 0:00 dovecot/imap-hibernate [18 connections] As you can see here I have 1258 IDLE imap processes (out of 2214 in total), and only 18 connections in imap-hibernate. I'd like to have all IDLE processes to be moved to imap-hibernate in order to save up memory. Any idea why these 1k+ IDLE imap processes are not moved into imap-hibernate? Thank you!

2 1

Chrooting
by Alexey Vlasov 30 Jun '23

30 Jun '23

Hi, I have the following hier for chrooting local users: /var/mail/.u336084 /var/mail/.u336084/bin /var/mail/.u336084/dev [cut here] /var/mail/.u336084/proc /var/mail/.u336084/var/ [cut here] /var/mail/.u336084/var/mail/ /var/mail/.u336084/var/mail/u336084 /var/mail/.u336084/var/mail/u336084/Mail /var/mail/.u336084/Mail -> var/mail/u336084/Mail (symlink) passwd file: u336084:x:1340334:255::/var/mail/.u336084:/bin/bash Chroot works fine, for SSH, FTP and etc, for example: # chroot /var/mail/.u336084 /bin/bash Given the current hierarchy, is it possible to tell the dovecot to do the same chroot? I found strange behavior that users cannot delete IMAP-folders: [pid 25055] stat("/var/mail/.u336084/Mail/.www1", {st_mode=S_IFDIR|0700, st_size=167, ...}) = 0 [pid 25055] stat("/var/mail/.u336084/Mail/.Trash.www1", 0x7ffeddc3d5c0) = -1 ENOENT (No such file or directory) [pid 25055] stat("/var/mail/.u336084/Mail/.Trash", {st_mode=S_IFDIR|0777, st_size=198, ...}) = 0 [pid 25055] fcntl(16, F_SETLK, {l_type=F_UNLCK, l_whence=SEEK_SET, l_start=0, l_len=0}) = 0 [pid 25055] fcntl(16, F_SETLK, {l_type=F_WRLCK, l_whence=SEEK_SET, l_start=0, l_len=0}) = 0 [pid 25055] fstat(16, {st_mode=S_IFREG|0600, st_size=0, ...}) = 0 [pid 25055] stat("/var/mail/.u336084/Mail/mailboxes.lock", {st_mode=S_IFREG|0600, st_size=0, ...}) = 0 [pid 25055] unlink("/var/mail/.u336084/Mail/mailboxes.lock") = 0 [pid 25055] close(16) = 0 [pid 25055] write(7, "96 NO [CANNOT] Renaming not supported across conflicting directory permissions (0.002 + 0.000 + 0.001 secs).\r\n", 110) = 110 [pid 25055] epoll_wait(11, [{EPOLLIN, {u32=1794732256, u64=94916277006560}}], 7, 1800000) = 1 Btw I don't see any conflicting dirs: root@:/var/mail/u336084# ls -ld /var/mail/.u336084/Mail/.www1 drwx------ 5 u336084 virtmail 167 Jun 27 13:26 /var/mail/.u336084/Mail/.www1 root@:/var/mail/u336084# ls -ld /var/mail/.u336084/Mail lrwxrwxrwx 1 u336084 virtmail 26 Apr 28 21:43 /var/mail/.u336084/Mail -> var/mail/u336084/Mail root@:/var/mail/u336084# ls -ld Mail drwx------ 16 u336084 virtmail 4096 Jun 30 01:25 Mail All folders have right owner u336084. With this configuration Courier worked properly, I noteced the problem after migration to Dovecot. Thanks in advance.

1 0

Warning sqlpool(mysql): Query failed
by Alessio Cecchi 27 Jun '23

27 Jun '23

Hi, with the last dovecot version (2.3.13) I see many of this warnings via "doveadm log errors": Feb 17 03:26:28 Warning: dict(31513): conn unix:dict (pid=31330,uid=95): dict(mysql)<email1(a)company1.com>: sqlpool(mysql): Query failed, retrying: MySQL server has gone away (idled for 679 secs) Feb 17 03:48:53 Warning: dict(31013): conn unix:dict (pid=31805,uid=95): dict(mysql)<email2(a)company2.com>: sqlpool(mysql): Query failed, retrying: MySQL server has gone away (idled for 685 secs) Feb 17 03:54:12 Warning: dict(32529): conn unix:dict (pid=3147,uid=95): dict(mysql)<email3(a)company3.com>: sqlpool(mysql): Query failed, retrying: MySQL server has gone away (idled for 670 secs) What do they mean? I thinks are related to inactivity timeout on mysql server (where I have a wait_timeout=600), but these warnings could be problems or can be fixed with some setting in dovecot? My dovecot configuration about dict: dict { acl = mysql:/etc/dovecot/dovecot-dict-sql.conf.ext expire = mysql:/etc/dovecot/dovecot-dict-sql.conf.ext sieve = mysql:/etc/dovecot/dovecot-dict-sieve-sql.conf.ext sqlquota = mysql:/etc/dovecot/dovecot-dict-sql.conf.ext } plugin { [...] acl_shared_dict = proxy::acl expire_dict = proxy::expire quota2 = dict:Quota Usage::noenforcing:proxy::sqlquota sieve = file:~/sieve;active=~/.dovecot.sieve sieve_before = dict:proxy::sieve;name=activesql [...] } service dict { process_limit = 500 unix_listener dict { group = vmail mode = 0660 user = vmail } } Thanks -- Alessio Cecchi Postmaster @ http://www.qboxmail.it https://www.linkedin.com/in/alessice

2 1

Re: Postfix: running a script on authentication failure
by Joseph Tam 27 Jun '23

27 Jun '23

On Thu, 22 Jun 2023, Michael Peddemors wrote: > * Use services like RATS-AUTH to block IPs that can safely be blocked as known hackers.. Cool. Are there other DNSRBLs (apart from bl.blocklist.de) that list BFD attack IPs? > * Use services like RATS-NULL (or SpamHaus DROP lists) right in the firewall level. There are SOME networks > that should simply be 'unplugged' Can't find it in https://spamrats.com/. Is it an DNSRBL or downloadable file? > * Turn off port 110 (well, all plain text authentication) 90% less email compromise reports when you do.. That will disable STARTTLS though. Even though it's not plaintext, maybe that is a good thing as it avoids MITM banner stripping attacks. Joseph Tam <jtam.home(a)gmail.com>

2 1

Server migration, password scheme/hashing, argon2i, argon2d, argon2id, sha512, sha512-crypt, tiger2, salt?
by David Mehler 24 Jun '23

24 Jun '23

Hello, I'm migrating to a new server. It's running Debian 11 currently though that's going 12 this weekend. Currently it uses Openssl v3.0.9, and dovecot 2.3.13 and MySQL (in this case Mariadb) for storing user account information v10.6.14. My question is in regards password storage and scheme/encryption/salts. Currently they are stored in Mariadb password field with a type of varchar and a 255 character length, and are stored as SHA512-CRYPT. I'm wondering if I should keep this as is or when I migrate go to another scheme? I'm thinking argon2i, argon2d, argon2id, sha512, sha512-crypt, tiger2, saltt? Discussion/opinions welcome. Thanks. Dave.

2 3

configuring Dovecot submission service
by Computerisms Corporation 23 Jun '23

23 Jun '23

Hi Dovecot Gurus, I have been playing around with the submission service. I have managed to get all the things working that I want working with two exceptions. I have not found (or possibly I have overlooked) solutions in the documentation or discussions on the 'net. First; I would like to be able to authenticate against my backend MTA using the user-supplied credentials. I have tried setting submission_relay_user as %u and submission_relay_password as %w, but these seem to be passed through as literal strings, and of course my backend MTA has no idea who %u is. I found a post from 2019 that indicates passing the submission_relay_user/password to the backend is not a supported function, but hoping that maybe that is outdated info. I also found: https://doc.dovecot.org/3.0/configuration_manual/authentication/proxies/ Which seems to indicate it could pass credentials to a non-Dovecot MTA backend, but I am not really clear from the docs on how to set that up, and I did not find any relevant examples to work from. Seems like it would be setting up a Director service, maybe? I have done that before and I do manage multiple email servers, but none of them is clustered or configured to be interoperable, so not sure if this is a viable approach. Second; I am wondering about the possibility of having multiple submission_relay_host entries. For example if I have dovecot submission services running on my standard port 587 and 465 connect to a relay on the same box through localhost, but could I have another submission service running on another port, say 5587, and relay all traffic received on that port through a completely different host? I am thinking I could set that up by running a separate instance of dovecot for just that service, but maybe there is a more suitable way? I don't have a solid use-case for this in mind at the moment, but given that I do manage multiple email servers, I am thinking that if this is possible it might provide some useful corner-case configurations down the road. Thank you for reading my post, any thoughts, information, or pointers to relevant documentation would be most appreciated. -- Bob Miller 867-334-7117 www.computerisms.ca

2 3

Postfix: running a script on authentication failure
by André Rodier 22 Jun '23

22 Jun '23

Hello, all. I just set-up a new server, running postfix, with submission(s) activated on standard ports (587, 465) Shortly after it has been setup, I see brute force attacks (not surprising) from a whole /24 network (more surprising). I carefully checked the logs, and see the modus operandi, which basically loop across the IP addresses in the network, to avoid being blacklisted by tools like fail2ban. And it is true, even with fail2ban activated, no IP is blacklisted. By activating verbose logging, I see multiple user names are tried, not only passwords. Is there any way, with postfix, to run a script on authentication failure, with information like the IP address and the username passed, for instance. I basically need features that fail2ban doesn't offer - I would like to not rely on reading logs, removing one step and acting more pro-actively. - If a script is called on authentication failure, it is fairly easy to use a Levenshtein distance to differentiate between a user having lost his password and a brute force attack. - If I log all the failure in a database, with the IP address, and the whois information, the script would take decision according to the whois information. What are you using on your side ? - Do you know any service, that I could use, to get the network to ban from an IP address reputation, something like crowdsec, for instance ? - Anyone has success with Suricata, Snort, or a tool like this ? Please, do not suggest third party hosted services, I want to be part of my self-hosting solution. Kind regards, André

4 4

Postfix: running a script on authentication failure
by André Rodier 22 Jun '23

22 Jun '23

Hello, all. I just set-up a new server, running postfix, with submission(s) activated on standard ports (587, 465) Shortly after it has been setup, I see brute force attacks (not surprising) from a whole /24 network (more surprising). I carefully checked the logs, and see the modus operandi, which basically loop across the IP addresses in the network, to avoid being blacklisted by tools like fail2ban. And it is true, even with fail2ban activated, no IP is blacklisted. By activating verbose logging, I see multiple user names are tried, not only passwords. Is there any way, with postfix, to run a script on authentication failure, with information like the IP address and the username passed, for instance. I basically need features that fail2ban doesn't offer - I would like to not rely on reading logs, removing one step and acting more pro-actively. - If a script is called on authentication failure, it is fairly easy to use a Levenshtein distance to differentiate between a user having lost his password and a brute force attack. - If I log all the failure in a database, with the IP address, and the whois information, the script would take decision according to the whois information. What are you using on your side ? - Do you know any service, that I could use, to get the network to ban from an IP address reputation, something like crowdsec, for instance ? - Anyone has success with Suricata, Snort, or a tool like this ? Please, do not suggest third party hosted services, I want to be part of my self-hosting solution. Kind regards, André

1 0

el9 rpms
by Marc 21 Jun '23

21 Jun '23

Is there a specific reason why there are (still) no el9 packages? https://repo.dovecot.org/ce-2.3.17/centos/ https://repo.dovecot.org/ce-2.3-latest/rhel/

4 7

Re: Submission behaviour
by Joseph Tam 20 Jun '23

20 Jun '23

From: "André Rodier" > chain input { > > # Limit new imap connections ala fail2ban > meta nfproto ipv4 tcp dport imaps ct state new,untracked \ > limit rate over 10/minute add @banned_imap_ipv4 { ip saddr } I'm don't know all the subttlties of this rule, but there are some mail clients (MacOSX Mail comes to mind) that will bombard your IMAP server with new connections when it does a global search. It will open a new connection for each mailbox, then do a search. When your connection limit is reached, it will then close all the open connections and do another round. This may be interpreted as a BFD attack, and you'll lock out a legitimate user. Joseph Tam <jtam.home(a)gmail.com>

1 0