1. Please take a little time for this simple survey! Thank you for participating!
    Dismiss Notice
  2. Dear Pleskians, please read this carefully! New attachments and other rules Thank you!
    Dismiss Notice
  3. Dear Pleskians, I really hope that you will share your opinion in this Special topic for chatter about Plesk in the Clouds. Thank you!
    Dismiss Notice

something is killing user processes

Discussion in 'Plesk 9.x for Linux Issues, Fixes, How-To' started by DariuszR, Jul 14, 2009.

  1. DariuszR

    DariuszR New Pleskian

    19
    85%
    Joined:
    Jul 14, 2009
    Messages:
    17
    Likes Received:
    0
    Hello

    I'm running Plesk 9.2.1 on CentOS 5.3. The problem is: each night at 00:36 something is killing user processes left running on screen or started as daemon (psyBNC for example). The only thing that I found in cron starting at 00:36 is "/usr/local/psa/admin/sbin/backupmng" but when I run it manually it doesn't kill anything. There are also strange entries in secure syslog at 00:36: usermod[pid]: lock user `username' password and usermod[pid]: unlock user `username' password for each ftp or shell user in plesk database. But again when I run it manually for user running some processes nothing is killed. Can somebody tell me what's going on here ?
     
  2. Webroy

    Webroy Regular Pleskian

    26
    23%
    Joined:
    Oct 29, 2008
    Messages:
    156
    Likes Received:
    0
    see if this is running: by ps -aux

    /usr/local/apache/bin/httpd -DSSL

    If so you have a problem , probably a not upgraded site is hacked...

    Search for hidden files like

    images/.thumbs.db/psybnc
    psybnc/bot.pl

    Update the scripts on that site and kill all httpd -DSSL processes....
     
  3. DariuszR

    DariuszR New Pleskian

    19
    85%
    Joined:
    Jul 14, 2009
    Messages:
    17
    Likes Received:
    0
    Server is freshly installed from OVH repos and nothing is hacked. I had fedora 4 with plesk before but I switched to CentOS 5.3 with completly fresh install and problems started right after that. All processes are killed, not psybnc only. You can run whatever you want, even just screen without any program running and it will be killed at 00:36. It looks like some maintenance plesk script but I can't find where is it and how to fix it.
     
  4. EduardoSilvestre

    EduardoSilvestre New Pleskian

    24
     
    Joined:
    Nov 10, 2008
    Messages:
    15
    Likes Received:
    0
    Location:
    Lisbon, Portugal
    Hello DariuszR,

    could be a fork caused by lack of memory. Check logs and looking for "Out of Memory".

    Regards,
     
  5. DariuszR

    DariuszR New Pleskian

    19
    85%
    Joined:
    Jul 14, 2009
    Messages:
    17
    Likes Received:
    0
    Now I'm 100% sure that something from Plesk is killing those processes. I have created user account directly from root not Plesk and left screen and psyBNC running over the night. Everything is still running fine and there is no "usermod" log entry for that user on 00:36. So can enybody tell me which Plesk script is doing this usermod changes each night ?
     
  6. DariuszR

    DariuszR New Pleskian

    19
    85%
    Joined:
    Jul 14, 2009
    Messages:
    17
    Likes Received:
    0
    It's not that. It's something from Plesk for sure (check my previous reply). Server has 4GB ram and usually ~3GB are free. Processes are killed at 00:36 every time so it's nothing random.
     
  7. TRANCEMANIAC

    TRANCEMANIAC Guest

    0
     
    Crontab of the root

    Hi DariuszR,

    you should ckeck the crontab of the root.

    Code:
    crontab -u root -l
    In my crontab i have this crons:
    Code:
    0       5       *       *       1       /opt/psa/libexec/modules/watchdog/cp/secur-check
    0       5       *       *       1       /opt/psa/libexec/modules/watchdog/cp/send-report weekly
    10      5       *       *       *       /opt/psa/libexec/modules/watchdog/cp/clean-sysstats
    15      5       *       *       *       /opt/psa/libexec/modules/watchdog/cp/pack-sysstats day
    15      5       *       *       1       /opt/psa/libexec/modules/watchdog/cp/pack-sysstats week
    15      5       1       *       *       /opt/psa/libexec/modules/watchdog/cp/pack-sysstats month
    15      5       1       *       *       /opt/psa/libexec/modules/watchdog/cp/pack-sysstats year
    20      5       *       *       *       /opt/psa/libexec/modules/watchdog/cp/clean-events
    0       5       *       *       7       /opt/psa/libexec/modules/watchdog/cp/clean-reports
    
    I have changed the hour from 1 to 5 and the restart of the apache comes every at about 5:00am - 5:20 am. So i think one of this crons is causing it.

    Also if you have a running backup task, it could be, that the apache locks the domain / complete apache for the time the backup takes.

    Hope it helps you.

    TRANCEY
     
Loading...