Update, 9 October 2026: the server-wide check now counts failures per account and per address over a fixed hour. It no longer treats a rising total of auth failed lines as proof of a hash problem or a reason for a bulk password reset. Found in an independent review.
Applies to cPanel & WHM 132 and later (Dovecot 2.4) on AlmaLinux 8/9 and CloudLinux 8/9; the /var/log/maillog path is for RHEL-family systems
Short answer: cPanel & WHM 132 moved to Dovecot 2.4, which blocks weak password hashes such as MD5-crypt ($1$) unless weak schemes are allowed, so mailboxes with old hashes fail as if the password were wrong. Read the failure reason in the log with grep -i 'auth failed' /var/log/maillog | tail -20. Fix hash problems with a password reset (UAPI Email passwd_pop), using Allow Weak Authentication Schemes only as a short bridge; fix “no auth attempts” by moving old clients to TLS 1.2 on port 993. (Reviewed October 2026.)
cPanel introduced Dovecot 2.4 with version 132 and made it standard across 134, 136 and 138, at the same time as tightening the hash algorithm used for mail account passwords. On most servers the upgrade is invisible. On a minority, mailboxes that had not changed their password in years, or clients using obsolete TLS or authentication mechanisms, stop logging in immediately after upcp. The symptoms look identical from the customer’s side (“wrong password”), so the first job is to separate the causes from the server log.
Table of Contents
Read the Dovecot log for the real reason
Every failed IMAP or POP login writes a line to /var/log/maillog with the reason after auth failed. Pull the last few hours for the affected account:
grep -i 'auth failed' /var/log/maillog | grep 'user@example.com' | tail -20
grep -i 'dovecot' /var/log/maillog | grep -iE 'ssl|tls|proto' | tail -20
The distinguishing phrases are:
unknown password schemeorPassword data is not valid— the stored hash is in a format the new Dovecot cannot verify.no auth attempts in 0 secsordisconnected (no auth attempts)— the client never sent credentials, which nearly always means a TLS handshake failure.Unsupported authentication mechanism— the client is asking for a mechanism that is now disabled.- Plain
(auth failed, 1 attempts in 2 secs): user=<...>, method=PLAINwith the correct password — the hash exists but does not verify.
The first and last are password-hash problems; the middle two are client or protocol problems.
Password-hash problems
cPanel stores mail passwords in /home/USER/etc/example.com/shadow. Older accounts may have hashes prefixed $1$ (MD5-crypt) or, on very old migrations, plain DES. The 2.4 packaging in cPanel prefers $6$ (SHA-512) and can be configured to refuse weak schemes. Inspect the stored hash prefix:
cut -d: -f1,2 /home/USER/etc/example.com/shadow | awk -F: '{print $1, substr($2,1,3)}'
Anything not starting with $6$ (or $2y$ for bcrypt on newer builds) is a candidate. The fix is a password reset, which rewrites the hash in the current scheme. You can do this without knowing the old password, either per account through cPanel » Email Accounts or in bulk from the shell:
uapi --user=USERNAME Email passwd_pop email=user domain=example.com password='NewStrongPassword'
For a large server, generate a list of all accounts with weak schemes first, then decide whether to reset them with notification or ask customers to reset themselves before a deadline. Resetting silently generates a support storm; announce it.
If you need a stopgap while customers reset, set Allow Weak Authentication Schemes to Yes in WHM » Service Configuration » Mailserver Configuration (added in version 132; check its current value first, because a 132 build changed its default). Enabling it lowers security and should be temporary; treat it as a bridge for a week, not a permanent state.
Client TLS and cipher failures
Dovecot 2.4 in cPanel disables TLS 1.0 and 1.1 and drops several weak ciphers. Clients that fail with “no auth attempts” are typically:
- Outlook 2010 or older on Windows 7 with no TLS 1.2 support.
- Old Android mail apps and some multifunction printers or scanners that send to IMAP.
- Anything configured for SSL/TLS on port 143 rather than STARTTLS, or STARTTLS on 993.
Confirm the server side is healthy first:
openssl s_client -connect mail.example.com:993 -tls1_2 </dev/null 2>&1 | grep -E 'Protocol|Cipher'
doveconf -n | grep -E 'ssl_min_protocol|ssl_cipher'
If the server negotiates TLS 1.2 fine, the fix is on the client: update the software, or change the client to port 993 with SSL/TLS and normal password authentication. Re-enabling TLS 1.0 server-wide to accommodate one printer is the wrong trade; if the device really cannot be replaced, an SMTP-only path on port 587 for outbound scans is usually enough.
Authentication mechanism mismatches
Some clients were configured years ago for CRAM-MD5 or NTLM. Dovecot 2.4 in cPanel enables PLAIN and LOGIN over TLS by default. Check the enabled list:
doveconf auth_mechanisms
If a customer’s client insists on an encrypted-password mechanism, switch the client to “normal password” (PLAIN over TLS is fully encrypted on the wire). Adding CRAM-MD5 server-side requires plaintext-recoverable password storage, which cPanel does not support, so this is not a server-side fix.
Sieve and Roundcube side-effects
Because 2.4 arrived with Pigeonhole Sieve integration in Roundcube, an upgrade also enables the managesieve service on port 4190. If a customer’s login works in a mail client but Roundcube shows a filter error, it is a different problem covered in Setting up Dovecot Sieve filters. Webmail login loops themselves are usually a Roundcube database issue, covered in our Roundcube guide.
A common pitfall: mixed cPanel builds behind one MX
Providers with an inbound mail proxy or a hot spare that still runs Dovecot 2.3 will see intermittent logins fail only for some users. The affected users are those whose passwords were reset on the 2.4 box in a scheme the 2.3 box cannot verify. Keep all servers that share mailbox storage or password files on the same major version; do not run a 2.3 secondary against 2.4 hashes.
Verify
For each fixed account, test directly rather than relying on the customer’s client:
curl -s --url 'imaps://mail.example.com/INBOX' --user 'user@example.com:NewStrongPassword' -X 'STATUS INBOX (MESSAGES)'
A * STATUS line back means authentication and TLS both work. Then re-check the log for new auth failed lines for that user over the following hour. To see the whole server, count failures per account over a fixed window (here the last hour) instead of a running total:
journalctl -u dovecot --since '-1 hour' --no-pager | grep 'auth failed' \
| grep -o 'user=<[^>]*>' | sort | uniq -c | sort -rn | head -20
journalctl -u dovecot --since '-1 hour' --no-pager | grep 'auth failed' \
| grep -o 'rip=[^,]*' | sort | uniq -c | sort -rn | head -10
Where rsyslog writes /var/log/maillog, the same lines are there too. These counts do not tell you why a login failed. A wrong password, a password-guessing attack (many accounts tried from a few rip= addresses) and an IP blocked by cPHulk all log the same auth failed line as an old hash. On our cPanel 138 lab server with Dovecot 2.4.5, a plain wrong-password login logged Login aborted: Connection closed (auth failed, 1 attempts in 2 secs) (auth_failed): user=<...>. Reset only accounts whose stored hash still uses a weak scheme: check each failing account’s hash prefix as shown above, and use the weak-scheme list from earlier for a planned reset. A rising count alone is not a reason for a bulk password reset. Keep an eye on whmapi1 version after the next update as well; Dovecot minor versions (2.4.5 shipped with 138) occasionally adjust defaults, and the changelog for your build will say if anything in the auth path moved again.
Official documentation (checked 9 October 2026): cPanel & WHM Release Notes · cPanel & WHM 132 Change Log · Mailserver Configuration (WHM documentation) · Password Schemes (Dovecot 2.4.2) · Upgrading Dovecot CE from 2.3 to 2.4 · doveconf(1) (Dovecot documentation) · Dovecot 2.4 settings reference · What are the minimum SSL protocol versions that Dovecot 2.4 accepts? (cPanel Support) · Where is email data stored on my cPanel server? (cPanel Support) · How To Reset E-mail User Passwords via the command line (cPanel Support) · Exim specification: The dovecot authenticator
Related guides: Exim outbound mail limits in WHM: hourly caps, X-Source tracking and “nobody” restrictions that stop spam runs · Warm up a new mail server IP or sending domain without landing in spam · Setting up MailBaby with cPanel/WHM Exim: router, transport, SRS and DKIM.
Frequently asked questions
Does the Dovecot 2.4 password-hash change also affect SMTP authentication in Exim?
Yes. Exim on cPanel authenticates senders through Dovecot’s auth socket, so a mailbox whose hash cannot be verified fails on port 587 as well as on IMAP and POP, and the same password reset fixes both.
How long does a bulk mail password reset take on a cPanel server?
The uapi call takes well under a second per account, so even thousands of mailboxes reset in minutes; the time that matters is the customer notification, so announce the deadline before running the loop.
Can I undo this?
A password reset cannot be reversed to the old hash, but customers can choose a new password of their own immediately afterwards; any temporary legacy-scheme setting enabled in Mailserver Configuration can be switched off again once resets are complete.
