• If you are still using CentOS 7.9, it's time to convert to Alma 8 with the free centos2alma tool by Plesk or Plesk Migrator. Please let us know your experiences or concerns in this thread:
    CentOS2Alma discussion

Issue Overnight we lost 20 users

Bjorn

Basic Pleskian
Hi,

Operating system: ‪CentOS Linux 7.7.1908 (Core)‬
Current used Plesk- Product: Plesk Onyx 17.8.11 Update #71
Depending installed mail - server software: Postfix

Production server that hosts 200 domains.

Yesterday (monday) morning we noticed a few domains returned a 503. After running the Diagnose & Repair most of the server went down and we had to reboot it.

After the restart some domains came back up, but still many websites returned a 503. After checking some files I noticed '/etc/passwd' was missing +/- 20 users. The webserver configuration troubleshooter showed many vhost conf creation errors.

When I checked all the '/vhosts/example.com' file permissions I noticed dir owners were mixed up. Domains without a user (missing ones) had a dir username similar to 10235. So I think somehow we lost 20 (linux/subscription) users and after running the 'File system repair', things got really messed up.

I did the following steps to get the domain back online.
1. adduser missing-user-1
2. Set correct home path & permission in /etc/passwd
(I know I could have done this in step1, but wanted to make sure skeleton structure did not go over.)
3. Chown the correct user for the 'vhosts/example.com' dir and childs.
4. Rebuild domainVhost config file.
5. Resave PHP settings for domain

After these steps the domain was live again. Other functions like creating a mail-account and in and outgoing mail is still working.

I got the following Questions.

1.
How did we lose +/- 20 linux users? Please advice which logs should I check to get an idea.

2.
During the missing users, we have disabled & activated one domain in the Plesk panel. Now this domain has mailing problems. When I create a new mail account I get the following error: "Error: mailmng-outgoing failed: ERROR: outgoing:mails.domain_id may not be NULL".
After running "plesk repair mail example.com" I see the same error.

We have the same symptoms you can see in this KB: Unable to create an additional domain: mailmng-outgoing failed.
After running "plesk repair mail -y" all restores return OK except "mail_outgoing_restore". And the notice: "Some utilities have exited with errors: /usr/lib64/plesk-9.0/remote_mail_restore/mail_outgoing_restore".

Please advice what steps I should follow to fix this.

3.
Is there a repair function to reset the correct permissions for users. The newly added users seem to have full root access with ftp. Permissions for the other users (not new) are still ok.

Note.
- It seems the psa DB is fine. Running "plesk repair db" returns all OK.
- SSH access is blocked by firewall.
- The server is owned by a small designer company, customers do not have access to FTP or panel.

Thank you in advance!
 
Last edited:
Item 2 is resolved.
"plesk repair mail -y" returns ok with 0 warnings or errors.
The domain with mailing problems is fixed.
 
Back
Top