hi all... i started this a few weeks ago.. the situation hasn't been improving... i'm getting really frustrated. i've looked for scripts but i can not find anything wrong. besides there are about 150 domains on that machine and i can't go through each php or pearl file.... i woke up today to about 20 or so stalled processes - the most ever. i do have server-status on but that just points to the machine name... what to do? i can't shut down domains until i figured where the hell is the offending script.... right now i have one process running wild but top and server-status report differnte cpu usage for those processes. here: from top: 55 processes: 2 running, 53 sleeping CPU states: 16.0% user, 0.0% nice, 83.7% system, 0.4% interrupt, 0.0% idle Mem: 349M Active, 428M Inact, 177M Wired, 38M Cache, 112M Buf, 9944K Free Swap: 2032M Total, 100K Used, 2032M Free PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND 66696 nobody 59 0 16168K 10580K RUN 153:54 98.24% 98.24% httpd 36782 qtss 2 0 6324K 4980K poll 59:51 0.00% 0.00% DarwinStr from server-status: Srv PID Acc M CPU SS Req Conn Child Slot Host VHost Request 0-0 66687 0/4206/4206 _ 7.34 3 22 0.0 114.66 114.66 65.214.44.36 richardedwardsnyc.com GET /s_s_singles.w.02/09.html HTTP/1.0 1-0 66688 0/4341/4341 _ 9.20 0 0 0.0 111.73 111.73 24.149.136.172 fotofolio.com GET /images/blbackbttn.gif HTTP/1.1 2-0 66689 11/4136/4136 K 7.63 0 0 0.0 85.68 85.68 68.111.73.32 TimesSquare.com GET /cut03/images/homecut/solo/frame_r4_c6.gif HTTP/1.1 3-0 66690 1/4426/4426 K 9.09 10 0 0.0 108.92 108.92 24.185.235.98 oldcove.org GET /images/juniorsailing.gif HTTP/1.1 4-0 66691 1/4416/4416 K 9.58 10 0 0.0 108.34 108.34 24.185.235.98 oldcove.org GET /images/T-juniorsailing.gif HTTP/1.1 5-0 66692 1/4552/4552 K 7.52 10 0 0.0 83.19 83.19 24.185.235.98 oldcove.org GET /images/spacer.gif HTTP/1.1 6-0 66693 0/4286/4286 W 8.07 111 0 0.0 122.70 122.70 86.127.1.145 giganticmusic.com GET /home.htm HTTP/1.1 7-0 66694 1/4737/4737 K 9.10 25 7817 99.8 77.52 77.52 68.111.73.32 TimesSquare.com POST /cut03/allResultsT.php HTTP/1.1 8-0 66695 0/4251/4251 _ 9.89 11 303 0.0 143.56 143.56 209.212.80.98 giganticmusic.com GET /smoker.swf HTTP/1.1 9-0 66696 0/3151/3151 W 6.50 9007 0 0.0 62.23 62.23 207.214.244.4 guey.el.net GET / HTTP/1.0 10-0 70405 0/1286/3848 _ 2.93 12 0 0.0 26.16 102.67 209.212.80.98 giganticmusic.com GET /favicon.ico HTTP/1.1 11-0 71906 11/22/3243 K 0.00 0 0 0.0 0.08 87.98 68.111.73.32 TimesSquare.com GET /cut03/images/rank2.0.gif HTTP/1.1 12-0 71921 12/14/2890 K 0.00 0 0 0.0 0.02 42.80 68.111.73.32 TimesSquare.com GET /cut03/images/rank2.5.gif HTTP/1.1 13-0 71923 14/14/2392 W 0.00 0 0 257.7 0.25 67.16 209.212.80.98 giganticmusic.com GET /images/shows/kjhlkjh.gif HTTP/1.1 14-0 71924 5/5/2025 K 0.00 0 0 15.7 0.02 22.99 209.212.80.98 giganticmusic.com GET /images/artists/someaction/sacover.jpg HTTP/1.1 15-0 71925 2/2/1634 W 0.00 4 0 20.4 0.02 37.66 24.90.33.115 guey.el.net GET /server-status HTTP/1.1 etc.. i'd really appreciate some help... thanks.... > Hi, > > This looks like a script that is running out of control. We've had similar > things here, except that our problem was the script writer failed to close > the mysql connections. The user was waiting for data which never came, hit > cancel, then tried again. The scripts used pconnect calls. This brought > a > 30,000 Euro server to its knees several times by eating all the available > sockets, at least until we found out what the problem was. > > In your case, set the max execution time for scripts to something short like > 10 seconds. Consider also disabling ini_set because we also had problems > with users playing with our settings and stressing our server. Be sure that > all uri includes are disabled too (XSS attack problems). > > To find the exact script is difficult. Something to note is your CPU usage > has 65% system. Look for scripts that were uploaded around the times your > server started having problems, then try to run them is one possibility to > find out what's happening. Look in the databases for connections that are > open and idle. Also, if your server comes down off its high load, look in > the apache log files to see which scripts just completed. Unfortunately apache writes to the logs only after the script or data has been (un)successfully sent. > > Markus > > > > On Tuesday 07 June 2005 19:02, kalin mintchev wrote: >> hi all... >> for the last 4 - 5 days i noticed something weird on one of the freebsd boxes here. one httpd process -or sometimes 2 or 3 - take up all the cpu. >> and it looks like this: > 8<--- > > --------------------------------------------------------------------- The official User-To-User support forum of the Apache HTTP Server Project. > See <URL:http://httpd.apache.org/userslist.html> for more info. > To unsubscribe, e-mail: users-unsubscribe@xxxxxxxxxxxxxxxx > " from the digest: users-digest-unsubscribe@xxxxxxxxxxxxxxxx > For additional commands, e-mail: users-help@xxxxxxxxxxxxxxxx > > -- -- "In a world without walls or fences, who needs Windows and Gates?" --------------------------------------------------------------------- The official User-To-User support forum of the Apache HTTP Server Project. See <URL:http://httpd.apache.org/userslist.html> for more info. To unsubscribe, e-mail: users-unsubscribe@xxxxxxxxxxxxxxxx " from the digest: users-digest-unsubscribe@xxxxxxxxxxxxxxxx For additional commands, e-mail: users-help@xxxxxxxxxxxxxxxx
|