Getting a SuSE 10.0 / Plesk 8.0.1 VPS on HostEurope up and running

Blog Topics:

There are a number of issues that need to be fixed/acknowledged before the VPS will run well.

  1. Most of the scripts in /etc/cron.daily don't run because they don't have the eXecutable bit set.

  2. /etc/cron.daily/logrotate does run, but when it finds something to do it produces the following error messages in /var/log/messages:
    logrotate: ALERT exited abnormally with [1]
    logrotate: ^G/usr/bin/mysqladmin: connect to server at 'localhost' failed
    logrotate: error: 'Access denied for user 'root'@'localhost' (using password: NO)'
    logrotate: error: error running postrotate script

    To allow logrotate to complete successfully, put your Plesk password and username into /root/.my.cnf, according to /etc/logrotate.d/mysql. You'll have to manually update .my.cnf whenever you change the password of the Plesk account.

  3. Plesk stores user passwords in MySQL, and if you use the passwd command to change a password from the shell command line, it will try to update the password in the database as well. This results in the following error messages in /var/log/messages:
    passwd[]: Old password verification started for user root
    passwd[]: Plesk DB connection established successfully
    passwd[]: Querying SELECT password, type FROM psa.accounts as a, psa.sys_users AS s 
                       WHERE = s.account_id AND s.login='root'
    passwd[]: No user 'root' found
    passwd[]: _plesk_verify_password returned 10, overriding with 24
    passwd[]: _get_stacked_password(pamh, flags=8192, opts=8, type=6, &password)
    passwd[]: unable to get PLESK password type, assuming default plain
    passwd[]: Plesk DB connection established successfully
    passwd[]: Querying SELECT psa.sys_users.account_id FROM psa.sys_users 
                       WHERE psa.sys_users.login = 'root'
    passwd[]: No user 'root' found
    passwd[]: password changed - account=root, uid=0, by=0

    Here I changed the password of the root user, which is not in the database, because root is not managed by Plesk. These errors are ugly, but they can safely be ignored.

  4. After checking the list of running System Services in Virtuozzo, you'll find that the following errors have been added to /var/log/messages:
    checkproc: checkproc: Usage:  checkproc [-v] [-k] [-p pid_file] /full/path/to/program
    checkproc: checkproc: cannot stat /usr/X11R6/bin/xdm: No such file or directory

    I hate seeing unexplained and unexpected error messages and traced them down to the following two commands which are apparently executed by Virtuozzo:

    /etc/rc.d/boot.udev status
    /etc/rc.d/xdm status

    So, the errors are due to sloppy shell scripts and can safely be ignored.

  5. I wasn't able to start the Watchdog module in Plesk...
    Error: The monit service does not respond. Refer to SWsoft technical support for help.

    ... and the VPS went into an endless loop trying to start Watchdog every 5 minutes. Here's from HE's answer:

    monit: The control file '/usr/local/psa/etc/modules/watchdog/monitrc' must have permissions no more than -rwxrw---- (0760); right now permissions are -rwxrwx--- (0770).

    I removed the "wdc" and "mon" lines from /etc/inittab and reset the VPS to stop the endless loop, and after tightening the permissions on monitrc, I was able to start Watchdog successfully.

    Unfortunately, it creates 10 to 20 (stunnel) entries in the system log every hour — I hate stuff that unnecessarily pollutes the system log!

It has taken many weeks to iron out these glitches, but finally the VPS runs fine.