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

High system load (Apache) and MySQL problem

Discussion in 'Plesk for Linux - 8.x and Older' started by kerm1t, Mar 4, 2005.

  1. kerm1t

    kerm1t Guest

    0
     
    Hello,

    My box hosts around 25 websites, most of which are Php/MySQL driven. The traffic they recieve is fairly modest, and the system itself is fairly powerful (3 GHz, 512 MB). Unfortunately, the load is rather high:

    # uptime
    15:23:36 up 18 days, 18:50, 3 users, load average: 7.70, 8.15, 8.21

    It's at this level constantly, so it can't be a blip caused by a particularly resource intensive script.

    Apache seems to be the main culprit:

    # ps aux|grep httpd
    root 25963 0.0 2.0 21608 10596 ? S 00:29 0:00 /usr/sbin/httpd
    root 25971 0.0 2.0 21552 10536 ? S 00:29 0:00 /usr/sbin/httpd
    apache 4155 11.1 2.5 23424 12988 ? R 10:54 30:17 /usr/sbin/httpd
    apache 4156 10.1 2.9 25380 15068 ? R 10:54 27:36 /usr/sbin/httpd
    apache 4212 10.1 2.9 25712 15284 ? R 10:55 27:36 /usr/sbin/httpd
    apache 4766 11.4 2.4 23048 12784 ? R 11:00 30:24 /usr/sbin/httpd
    apache 4767 11.2 2.8 24768 14368 ? R 11:00 30:02 /usr/sbin/httpd
    apache 4872 11.5 3.0 25692 15364 ? R 11:00 30:34 /usr/sbin/httpd
    apache 4873 12.8 2.9 25476 15116 ? R 11:00 33:58 /usr/sbin/httpd
    apache 20915 0.6 2.7 24172 14044 ? S 13:43 0:40 /usr/sbin/httpd
    apache 20927 0.6 2.8 24432 14344 ? S 13:43 0:41 /usr/sbin/httpd
    apache 21090 0.6 2.7 24168 14092 ? S 13:45 0:39 /usr/sbin/httpd
    apache 22027 0.6 2.7 24316 14152 ? S 13:53 0:36 /usr/sbin/httpd
    apache 23973 0.5 2.7 24380 14300 ? S 14:12 0:25 /usr/sbin/httpd
    apache 24564 0.5 2.7 24176 14120 ? S 14:18 0:23 /usr/sbin/httpd
    apache 24790 0.5 2.6 23828 13752 ? S 14:22 0:22 /usr/sbin/httpd
    apache 24791 0.5 2.7 24068 14020 ? S 14:22 0:21 /usr/sbin/httpd
    apache 24792 0.5 2.7 24292 14252 ? S 14:22 0:20 /usr/sbin/httpd
    apache 27643 0.5 2.7 24052 13936 ? S 14:58 0:09 /usr/sbin/httpd
    apache 28475 0.5 2.7 23972 13868 ? S 15:08 0:05 /usr/sbin/httpd
    root 29758 0.0 0.1 3688 668 pts/2 S 15:25 0:00 grep httpd

    StartServers is defined in httpd.conf as 8. Presumably the PIDs 4xxx are these initial servers, and the higher PIDs have been spawned to deal with client requests.

    Any thoughts why these initial servers are using so much CPU time? How would I go about determining where the bottleneck is (ie is it a Php script, or perhaps a modrewrite)



    On a possibly related noted, I upgraded MySQL a couple of days ago (from 3.23 to 4.1). I've been starting it from it's init script in /etc/rc.d/init.d/ (the Plesk CP doesn't appear to offer a way to start/stop MySQL), but it's ignoring the pid-file setting in /etc/my.cnf, and writing it's PID to a different filename (was mysql.pid, now it's $hostname.pid).

    There are currently 45 instances of mysql running, and the number seems to go up by a few everyday (it doesn't seem to be related to the number of http processes, ie I don't believe all these mysql processes are being spawned by Apache serving up dynamic pages to visitors). Could the misnamed pid file be causing fresh MySQL processes to launch?

    Thanks
     
Loading...