High server load with nobody httpd process.

[xxingan]

Hi. I have a dedicated server , which with celeron 2.3G and 1G memory.

Recently it is always around 4-6 server load and sometimes over 10.

I checked in ssh "top -c" :

17411 nobody 17 0 68032 39m 4244 R 35.4 4.1 1:17.11 /usr/local/apache/bin/httpd -k start -DSSL
17482 nobody 17 0 68044 39m 4212 R 34.4 4.1 0:26.37 /usr/local/apache/bin/httpd -k start -DSSL
17684 nobody 17 0 67916 39m 4160 S 29.8 4.0 0:26.81 /usr/local/apache/bin/httpd -k start -DSSL
17791 root 15 0 1968 968 760 R 0.7 0.1 0:00.06

There are several httpd processes waste server resources . They always on the top. If I stop the apache , the server load will back to normal.

Tasks: 104 total, 16 running, 87 sleeping, 1 stopped, 0 zombie
Cpu(s): 15.2% us, 84.4% sy, 0.0% ni, 0.0% id, 0.0% wa, 0.0% hi, 0.3% si
Mem: 1001548k total, 988116k used, 13432k free, 81404k buffers
Swap: 2096440k total, 560k used, 2095880k free, 632632k cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
17865 nobody 17 0 67904 39m 4152 R 39.6 4.0 0:03.11 httpd
17688 nobody 17 0 68040 39m 4168 R 36.0 4.0 0:53.01 httpd
17881 nobody 17 0 67540 38m 3824 R 23.4 4.0 0:04.15 httpd
17883 root 15 0 1964 964 760 R 0.3 0.1 0:00.09 top

sometimes a php process also take some resources.


How can I fix this nobody httpd problem? My server host serveral web accounts but almost no traffic. So I believe there will be a problem ..

Any comments welcome.. help me..

THanks.

[AndyReed]

How can I fix this nobody httpd problem?

Re-compile Apache with SuPhp. BTW, this issue has been covered many times in these forums. Search?

[LiNUxG0d]

Hey there,

Firstly, check in your WHM under Server Status -> Apache Status to bring up what VHost is being accessed, and which files.

Secondly, look at your PIDs here:

17411 nobody 17 0 68032 39m 4244 R 35.4 4.1 1:17.11 /usr/local/apache/bin/httpd -k start -DSSL
17482 nobody 17 0 68044 39m 4212 R 34.4 4.1 0:26.37 /usr/local/apache/bin/httpd -k start -DSSL
17684 nobody 17 0 67916 39m 4160 S 29.8 4.0 0:26.81 /usr/local/apache/bin/httpd -k start -DSSL

The IDs are 17411, 17482 and 17684. Find them in the Apache Status section of WHM using the "Search" option in your browser and then you'll have a good idea of where to start looking for the resource hog.

Hope this helps you!

Let me know if you have any follow-up questions.

Warmest regards,

[LiNUxG0d]

Re-compile Apache with SuPhp. BTW, this issue has been covered many times in these forums. Search?

Hey AndyReed,

I think this particular problem isn't that SuPHP isn't installed and that httpd is running as nobody, it's more that his httpd is causing a high load. I don`t think there are any real benefits to installing suPHP in this specific instance, except maybe knowing which user is accessing specific PHP files; httpd will still run as nobody anyway.

Installing suPHP without proper guidance will simply cause problems for you right now. I think that AndyReed is right though, suPHP should be installed, but not to fix this specific issue.

If you have any follow-up questions or anything, let me know.

Warmest regards,

[xxingan]

Thanks Andy and LinuxG0d

I did recompile the apache with suphp mod . The problems are still there.

I tried to read from apache status and see almost all those processes are related to php.

Whatever the process is a wordpress xmlrpc or some php page, it keep high server load.

I tried to move same file to other server and seemed worked fine.



Any comments welcome ~

Leon

[Kent Brockman]

Thanks Andy and LinuxG0d

I did recompile the apache with suphp mod . The problems are still there.

I tried to read from apache status and see almost all those processes are related to php.

Whatever the process is a wordpress xmlrpc or some php page, it keep high server load.

I tried to move same file to other server and seemed worked fine.



Any comments welcome ~

Leon

Hi, did you checked the contents of "CPU/Memory/MySQL Usage" in WHM? this is under the "Server Status" section. check and see what domains have the highest usage of resources. Note that this is a daily report, showing the top resource users, so you can browse past days and see how this have behave You'll be able to narrow down the problem as you see what domain is causing the high load.

If you don't realize the results you see in that screen, feel free to paste them here to let's see and comment what may be happening.

Also, if you have SSH access, you can check if some error are hapening just right now, by checking contents of this log:

less /usr/local/apache/logs/error_log

you can check executed scripts in your hosted domains checking the following 2 logs:

less /usr/local/apache/logs/suphp_log
less /usr/local/apache/logs/suexec_log

[LiNUxG0d]

Thanks Andy and LinuxG0d

I did recompile the apache with suphp mod . The problems are still there.

I tried to read from apache status and see almost all those processes are related to php.

Whatever the process is a wordpress xmlrpc or some php page, it keep high server load.

I tried to move same file to other server and seemed worked fine.



Any comments welcome ~

Leon

Ahhh Leon,

Wondrous Wordpress is being a hog is it? Yeah, we have that happening lots on servers that we support and don't understand why it's happening. We have lots of "cron.php" files that seem to just pile up and not get processed... I don't know why.

I've seen people allude to some bugs with some scripts that they run, but because Wordpress is so modular, it could be a module that's being a bad apple. I keep hearing about this "Ping'o'matic" Wodpress plug-in that can be VERY problematic.

Take the process that's being a pest, find it in Apache Status and paste it here. Leave the URL untouched, but feel free to Mast the VHost if you want.

Let me know, I'm curious if we're affected by the same issue.

Regards,

[Kent Brockman]

Wondrous Wordpress is being a hog is it? Yeah, we have that happening lots on servers that we support and don't understand why it's happening. We have lots of "cron.php" files that seem to just pile up and not get processed... I don't know why.

I've seen people allude to some bugs with some scripts that they run, but because Wordpress is so modular, it could be a module that's being a bad apple. I keep hearing about this "Ping'o'matic" Wodpress plug-in that can be VERY problematic.

We also have some customers using Wordpress blogs. In fact, we decided to deploy our own wordpress blog to study the case and high resource usage seems to be a daily common thing.

When you see the results under "CPU/Memory/MySQL Usage" section, you can see that domains hosting Wordpress blogs reache peaks of even 25% of CPU. The main Load Balancing is not generally affected, or at less, we haven't blogs popular enough to generate big traffic.

By now, we are deploying Wordpress 2.3.x instead the brand new 2.5, because not all the MUST-HAVE plugins are totally compatible with 2.5. Till then, we'll not recommend the upgrade. However, I have readed that v2.5 manages resource and memory usage in a better way.

BTW, you can replace ping-o-matic queries by another ones:

http://rpc.weblogs.com/
http://ping.blo.gs/
http://ping.blogalaxia.com
http://api.feedster.com/ping
http://ping.bitacoras.com
http://rpc.technorati.com/rpc/ping
http://ping.feedburner.com
http://rpc.blogrolling.com/pinger/
http://rpc.newsgator.com/
http://api.my.yahoo.com/rss/ping

Also, you may accelerate the blog using the plugin "WP Super Cache" which has done an spectacular work. But won't lower significantly the CPU usage


But hey!!! this is not a Wordpress forum!!!!
Keep the focus boys!

[LiNUxG0d]

We also have some customers using Wordpress blogs. In fact, we decided to deploy our own wordpress blog to study the case and high resource usage seems to be a daily common thing.

When you see the results under "CPU/Memory/MySQL Usage" section, you can see that domains hosting Wordpress blogs reache peaks of even 25% of CPU. The main Load Balancing is not generally affected, or at less, we haven't blogs popular enough to generate big traffic.

By now, we are deploying Wordpress 2.3.x instead the brand new 2.5, because not all the MUST-HAVE plugins are totally compatible with 2.5. Till then, we'll not recommend the upgrade. However, I have readed that v2.5 manages resource and memory usage in a better way.

BTW, you can replace ping-o-matic queries by another ones:

http://rpc.weblogs.com/
http://ping.blo.gs/
http://ping.blogalaxia.com
http://api.feedster.com/ping
http://ping.bitacoras.com
http://rpc.technorati.com/rpc/ping
http://ping.feedburner.com
http://rpc.blogrolling.com/pinger/
http://rpc.newsgator.com/
http://api.my.yahoo.com/rss/ping

Also, you may accelerate the blog using the plugin "WP Super Cache" which has done an spectacular work. But won't lower significantly the CPU usage


But hey!!! this is not a Wordpress forum!!!!
Keep the focus boys!

Well, in this case Wordpress is becoming quite the focusing point here.

httpd is going high load because of Wordpress. So we're really can't not focus on it... right?

Hehehehehe...

I guess your options are few and far between then. To be or not to be, that's the question.

Suspend him or don't. Regardless, Wordpress is a memory hog.

OINK!

[xxingan]

Dear

Seemed the issue is related to wordpress update.

I'm talking with my customers. They also setup some cronjob for automatic update the blog.

I have no idea about how to lower the usage . Maybe move them to a high spec server instead this one.

Cheers..

[ramprage]

yes wordpress is a resource muncher at times, so keep an eye on it.

[123addone]

i have this kind issue now with one hosting and they tell me that i need to upgrade when i had same sites on 1gb ram server and i have 2gb ram now, what's can be wrong here ?

[cPanelDon]

To better organize discussion and increase relevant attention of the topic this thread has been relocated into the cPanel and WHM Optimization forums: cPanel and WHM Optimization - cPanel Forums

[wjdennen]

These problems sound familiar to this bug in Eaccelerator:

#368 (hung httpd threads with high page request rates) - eAccelerator - Trac

[Monie]

These problems sound familiar to this bug in Eaccelerator:

#368 (hung httpd threads with high page request rates) - eAccelerator - Trac

I'm going to try taking out eAccelerator and see how this turns out.

EDIT: Taking out eAccelerator has brought down the server load drastically and it hasn't spiked to the 90's yet.