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

8.2 upgrade causing high server loads every hour

Discussion in 'Plesk for Linux - 8.x and Older' started by danberlyoung, Oct 23, 2007.

  1. danberlyoung

    danberlyoung Guest

    I upgraded to 8.2 on 10/16/07 and ever since then several strange things started happening. First the Plesk Server and spamassassin services started shutting down and being restarted by watchdog nearly every hour. I checked the " Resources Usage Statistics" in the Watchdog module and I can see the cpu utilization peak (80-90%) every hour on the hour. It ramps up for about 15 minutes before then and then abruptly falls off after the top of the hour. I tried shutting down spamassassin thinking it might be the culprit but that did not make any difference.

    I looked at TOP for a little while and it seems that gzip is running a *lot* and seems to peak when the cpu utilization peaks. Also, the web server and admin interface get really sluggish during those peaks.

    Anyone else seeing this? Anyone have any ideas on how to see what is going on? Oh, I'm running Plesk 8.2 on CentOS 4.

    Thanks in advance!
  2. danberlyoung

    danberlyoung Guest

    After poking around and turning numerous things on and off I couldn't find much of anything. And the activity spikes changed to about every 15 minutes so I was completely flummoxed.

    Then in desperation, I tried a modification I found on line here that adjusted the /etc/xinet.d/smpt_psa file. (The article didn't mention this but I made the same edit to the smpts_psa file in the same directory.) I then restarted qmail and xinetd and POOF! Nothing happened.

    But then about 3 hours later, the cpu utilization (that was hovering between 40 and 80%) suddenly dropped to near 0 and stayed there. WTF? It's been there now for nearly 24 hours with only an occasional normal looking spike and everything seems OK still.

    And as a added benefit, one of my clients that was having difficulty receiving email from authorize.net suddenly started receiving them. Go figure.

    All's well that ends.
  3. jaredfine

    jaredfine Guest

    Weird, -Rt0 should really have nothing to do with the issue you were having. Basically it just means set the reverse DNS lookup timeout to 0, effectively disabling reverse lookup.

    I'm going to be upgrading to 8.2 shortly, presuming I don't read too much negative stuff on it. I'll post back if I have similar issues.

    Glad it worked for you though.
  4. dirty@

    dirty@ Guest

    every 15 minutes backupmng is run, it check if there are any scheduled backups (which can be set by clients and domain owners themselves) and if scheduled backup is found, then it is run..

    this also may describe why you see gzip in top - backups are really gzipped..

    run `ps axfww` or `pstree` during peaks to see all the processes run, it may help to understand the reason..