Facebook
Twittercreativaonline
New Pleskian
- Server operating system version
- AlmaLinux / RHEL 8
- Plesk version and microupdate number
- 18.0.76 update #5
OS: Linux (AlmaLinux / RHEL 8)
- Plesk version: 18.0.76, Update 5 (build 2026-04-06, revision f5323d07)
- Mail server: Dovecot (Plesk managed)
- MTA: Postfix
- Server type: Dedicated, ~80-100 hosted domains, several hundred mailboxes
---
SUMMARY
After every automatic Plesk update that includes plesk-mail-pc-driver and/or
psa-mail-driver-common, Dovecot loses its authentication database synchronization
with Plesk. Hundreds of mailboxes across all hosted domains become inaccessible,
returning "wrong username or password" errors to end users.
The issue repeats itself every night after the update, requiring manual execution
of "plesk repair mail -y" to restore service.
---
TIMELINE
2026-04-08 03:20h — Plesk automatic update ran and upgraded the following packages:
plesk-mail-pc-driver-18.0-2.redhat.8+p18.0.76.5+t260406.1134.x86_64
psa-mail-driver-common-18.0-2.redhat.8+p18.0.76.5+t260406.1134.x86_64
plesk-core-18.0-2.redhat.8+p18.0.76.5+t260406.1134.x86_64
plesk-control-panel-18.0-2.redhat.8+p18.0.76.5+t260406.1134.noarch
plesk-roundcube-1.6.15-2.redhat.8+p18.0.76.5+t260406.1134.noarch
(and others)
2026-04-08 03:29h — First Dovecot authentication failures appear in /var/log/maillog:
dovecot_authdb_plesk[...]: No such user '[email protected]' in mail authorization database
This error appeared for hundreds of mailboxes across 12+ domains within minutes.
2026-04-08 (daytime) — Issue diagnosed and resolved by running:
plesk repair mail -y
2026-04-09 03:53h — Issue recurred the following night. No Plesk mail-related packages
were updated on April 9th. Only cloudlinux-linksafe-1-5.2.el8 was upgraded.
The desync appears to happen again after nightly service restarts triggered by
cron jobs / system activity around 03:00-04:00h.
---
ERROR EVIDENCE
Sample log entries from /var/log/maillog:
Apr 8 03:30:40 dedi50173 dovecot_authdb_plesk[1729475]: No such user '[email protected]' in mail authorization database
Apr 8 03:33:17 dedi50173 dovecot_authdb_plesk[1729475]: No such user '[email protected]' in mail authorization database
Apr 8 04:09:30 dedi50173 dovecot_authdb_plesk[1729475]: No such user '[email protected]' in mail authorization database
Apr 8 05:59:45 dedi50173 dovecot_authdb_plesk[1729475]: No such user '[email protected]' in mail authorization database
...
(200+ similar entries across 12+ domains)
Apr 9 03:53:52 dedi50173 dovecot_authdb_plesk[4159833]: No such user '[email protected]' in mail authorization database
...
(Issue recurred the following night)
---
ADDITIONAL DATABASE INCONSISTENCY FOUND
Running "plesk repair db -y" revealed a related inconsistency:
Inconsistency in table 'DomainServices' for column ipCollectionId:
Web service cannot operate: the IP address list with ID 274 is empty
Inconsistency in table 'DomainServices' for column ipCollectionId:
Web service cannot operate: the IP address list with ID 348 is empty
Affected domains: nuevoalmacen.es (dom_id 180) and omics-clm.es (dom_id 141).
Both domains use the server only for mail (web hosted elsewhere), which left
their mail service records pointing to empty IP collections in DomainServices.
"plesk repair db" could not resolve this automatically.
Manual fix applied:
INSERT IGNORE INTO IpAddressesCollections (ipCollectionId, ipAddressId) VALUES (274, 1);
INSERT IGNORE INTO IpAddressesCollections (ipCollectionId, ipAddressId) VALUES (348, 1);
This fixed the repair db loop but the nightly Dovecot desync persists.
---
STEPS TO REPRODUCE
1. Have a Plesk server with 50+ domains and mailboxes
2. Allow automatic Plesk updates to run (includes plesk-mail-pc-driver)
3. After the update, Dovecot authentication database loses sync with Plesk psa DB
4. All mailbox users get "wrong username or password" despite correct credentials
5. Running "plesk repair mail -y" restores access temporarily
6. The desync recurs nightly after service restarts
---
WORKAROUND (temporary)
Scheduled a daily cron job at 03:15 AM to run "plesk repair mail -y" automatically:
15 3 * * * /usr/sbin/plesk repair mail -y >> /var/log/plesk-repair-mail.log 2>&1
This mitigates the impact but does not fix the root cause.
---
EXPECTED BEHAVIOR
After a Plesk update, Dovecot authentication database should remain in sync with
Plesk. Running "plesk repair mail" should not be required after every update or
every night to keep mailboxes accessible.
---
QUESTIONS
1. Is this a known bug in plesk-mail-pc-driver 18.0.76.5?
2. Is there a permanent fix or patch available?
3. Should we rollback plesk-mail-pc-driver to a previous version?
4. How can we prevent automatic updates from breaking Dovecot sync in the future?
---
- Plesk version: 18.0.76, Update 5 (build 2026-04-06, revision f5323d07)
- Mail server: Dovecot (Plesk managed)
- MTA: Postfix
- Server type: Dedicated, ~80-100 hosted domains, several hundred mailboxes
---
SUMMARY
After every automatic Plesk update that includes plesk-mail-pc-driver and/or
psa-mail-driver-common, Dovecot loses its authentication database synchronization
with Plesk. Hundreds of mailboxes across all hosted domains become inaccessible,
returning "wrong username or password" errors to end users.
The issue repeats itself every night after the update, requiring manual execution
of "plesk repair mail -y" to restore service.
---
TIMELINE
2026-04-08 03:20h — Plesk automatic update ran and upgraded the following packages:
plesk-mail-pc-driver-18.0-2.redhat.8+p18.0.76.5+t260406.1134.x86_64
psa-mail-driver-common-18.0-2.redhat.8+p18.0.76.5+t260406.1134.x86_64
plesk-core-18.0-2.redhat.8+p18.0.76.5+t260406.1134.x86_64
plesk-control-panel-18.0-2.redhat.8+p18.0.76.5+t260406.1134.noarch
plesk-roundcube-1.6.15-2.redhat.8+p18.0.76.5+t260406.1134.noarch
(and others)
2026-04-08 03:29h — First Dovecot authentication failures appear in /var/log/maillog:
dovecot_authdb_plesk[...]: No such user '[email protected]' in mail authorization database
This error appeared for hundreds of mailboxes across 12+ domains within minutes.
2026-04-08 (daytime) — Issue diagnosed and resolved by running:
plesk repair mail -y
2026-04-09 03:53h — Issue recurred the following night. No Plesk mail-related packages
were updated on April 9th. Only cloudlinux-linksafe-1-5.2.el8 was upgraded.
The desync appears to happen again after nightly service restarts triggered by
cron jobs / system activity around 03:00-04:00h.
---
ERROR EVIDENCE
Sample log entries from /var/log/maillog:
Apr 8 03:30:40 dedi50173 dovecot_authdb_plesk[1729475]: No such user '[email protected]' in mail authorization database
Apr 8 03:33:17 dedi50173 dovecot_authdb_plesk[1729475]: No such user '[email protected]' in mail authorization database
Apr 8 04:09:30 dedi50173 dovecot_authdb_plesk[1729475]: No such user '[email protected]' in mail authorization database
Apr 8 05:59:45 dedi50173 dovecot_authdb_plesk[1729475]: No such user '[email protected]' in mail authorization database
...
(200+ similar entries across 12+ domains)
Apr 9 03:53:52 dedi50173 dovecot_authdb_plesk[4159833]: No such user '[email protected]' in mail authorization database
...
(Issue recurred the following night)
---
ADDITIONAL DATABASE INCONSISTENCY FOUND
Running "plesk repair db -y" revealed a related inconsistency:
Inconsistency in table 'DomainServices' for column ipCollectionId:
Web service cannot operate: the IP address list with ID 274 is empty
Inconsistency in table 'DomainServices' for column ipCollectionId:
Web service cannot operate: the IP address list with ID 348 is empty
Affected domains: nuevoalmacen.es (dom_id 180) and omics-clm.es (dom_id 141).
Both domains use the server only for mail (web hosted elsewhere), which left
their mail service records pointing to empty IP collections in DomainServices.
"plesk repair db" could not resolve this automatically.
Manual fix applied:
INSERT IGNORE INTO IpAddressesCollections (ipCollectionId, ipAddressId) VALUES (274, 1);
INSERT IGNORE INTO IpAddressesCollections (ipCollectionId, ipAddressId) VALUES (348, 1);
This fixed the repair db loop but the nightly Dovecot desync persists.
---
STEPS TO REPRODUCE
1. Have a Plesk server with 50+ domains and mailboxes
2. Allow automatic Plesk updates to run (includes plesk-mail-pc-driver)
3. After the update, Dovecot authentication database loses sync with Plesk psa DB
4. All mailbox users get "wrong username or password" despite correct credentials
5. Running "plesk repair mail -y" restores access temporarily
6. The desync recurs nightly after service restarts
---
WORKAROUND (temporary)
Scheduled a daily cron job at 03:15 AM to run "plesk repair mail -y" automatically:
15 3 * * * /usr/sbin/plesk repair mail -y >> /var/log/plesk-repair-mail.log 2>&1
This mitigates the impact but does not fix the root cause.
---
EXPECTED BEHAVIOR
After a Plesk update, Dovecot authentication database should remain in sync with
Plesk. Running "plesk repair mail" should not be required after every update or
every night to keep mailboxes accessible.
---
QUESTIONS
1. Is this a known bug in plesk-mail-pc-driver 18.0.76.5?
2. Is there a permanent fix or patch available?
3. Should we rollback plesk-mail-pc-driver to a previous version?
4. How can we prevent automatic updates from breaking Dovecot sync in the future?
---
