0

I have a mail server with a few dozen IMAP accounts. The configuration is correct for each account, every one of them tests OK.

When a client from a particular, very slow Internet connection attempts to connect, I get messages like these:

Mar 14 07:58:31 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 2 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<ETMgqX9mDraoAyqn>
Mar 14 07:58:31 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 2 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<EPggqX9mfqGoAyqn>
Mar 14 07:58:31 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 2 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<E3kgqX9mYJioAyqn>
Mar 14 07:58:37 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 4 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<YCZXqX9mHZSoAyqn>
Mar 14 07:58:37 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 4 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<5/JcqX9miKaoAyqn>
Mar 14 07:58:37 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 4 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<HTJYqX9m1p+oAyqn>
Mar 14 08:00:11 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 3 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<pesGr39mrICoAyqn>
Mar 14 08:00:11 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 3 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<bJ0Gr39mqYqoAyqn>
Mar 14 08:00:11 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 3 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<La0Gr39muZ2oAyqn>
Mar 14 08:02:08 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 2 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<lQsHtn9mH62oAyqn>
Mar 14 08:02:08 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 2 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<V8sHtn9mXruoAyqn>
Mar 14 08:02:08 mailserv dovecot: imap-login: Aborted login (auth failed, 2 attempts in 2 secs): user=<>, method=LOGIN, rip=111.111.111.111, lip=222.222.222.222, TLS, session=<WSIHtn9m5beoAyqn>

The client is at 111.111.111.111 here, the server at 222.222.222.222.

The client can at the same time successfully query three accounts on the server. They do not know of any other accounts queried from their client machine.

The possibilities for the problem are either a second, forgotten mail program on the computer trying a connect unbeknownst to the user (who is not a computer person), or a timeout issue due to a very slow (modem dialup) Internet connection of the client machine.

My questions therefore:

  1. Can I find out which username and which client user agent (and perhaps even, which password) is used in these failed connection attempts? I have tried setting

    debug_log_path = /var/log/dovecot-debug.log
    auth_debug = yes
    

However, I could not find anything relevant in these (very noisy) logs, perhaps it is the wrong log file to record failed (or timed out) attempts? Why does the log say "user=<>"? Can I make that log line more helpful?

  1. Can I ask dovecot whether these were wrong passwords or perhaps timeouts? I have set

    auth_policy_server_timeout_msecs = 17000
    

(default 3000) to no avail - but on the other hand the logs do say "2 attempts in 2 secs" so can I perhaps exclude a timeout anyway?

4
  • Show your logs. Whether its dovecot or PAM handling the authentication, either way you should be getting entries very clearly differentiating between connection and credential problems. Note the auth_verbose_passwords=sha1 setting which allows you to log passwords in an obscured fashion, so you do not spill as much sensitive information into log files.
    – anx
    Mar 14, 2021 at 20:01
  • @anx Thanks, so I would combine auth_debug = yes with auth_verbose_passwords=sha1, right? Then what do I grep in that file (100k+ lines, busy server, 99% of logins without problems)? Can I add user agents to the log?
    – Ned64
    Mar 15, 2021 at 22:10
  • 1
    When these protocols were standardized, dialup was what everyone had. It seems unlikely that the connection speed or latency is a factor. Mar 15, 2021 at 22:24
  • @MichaelHampton OK, thanks. So how can I find out as much as I can about the attempt that fails?
    – Ned64
    Mar 16, 2021 at 18:38

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.