Server load issues

Portal Home > Knowledgebase > Articles Database > Server load issues

Posted by Woooo, 06-30-2008, 12:06 PM
Hi, One of my servers is recently having big loads issues, it goes upto 300 at particular times and bogs down everything. thr top header looks like it : top - 11:02:48 up 2 days, 6:27, 2 users, load average: 28.87, 61.16, 43.88 Tasks: 846 total, 1 running, 845 sleeping, 0 stopped, 0 zombie Cpu(s): 13.2% us, 20.6% sy, 0.0% ni, 64.3% id, 61.4% wa, 0.1% hi, 0.5% si Mem: 4147464k total, 3962484k used, 184980k free, 38416k buffers Swap: 4192944k total, 592k used, 4192352k free, 1798536k cached The %wa is always heading towards heights, What is it? is it disk input/output related thing ? or RAM related thing? CPU is always free at many momets when this problem comes. What should i tweak and check? will having data spread on a secondary disk will help with IP performance? Please guide. Thanks
Posted by Woooo, 06-30-2008, 12:07 PM
top - 11:06:36 up 2 days, 6:31, 2 users, load average: 5.90, 31.58, 35.55 Tasks: 976 total, 3 running, 971 sleeping, 0 stopped, 2 zombie Cpu(s): 15.4% us, 35.0% sy, 0.0% ni, 48.3% id, 0.7% wa, 0.1% hi, 0.6% si Mem: 4147464k total, 3983872k used, 163592k free, 43464k buffers Swap: 4192944k total, 592k used, 4192352k free, 1892620k cached
Posted by The Universes, 06-30-2008, 12:20 PM
wa is iowait, and its really high. What are you running on the system? (cpu intensive, database, dynamic site, etc?) Do you have systat installed? If you do, run iostat and see how many reads/writes are occurring when there is high iowait.
Posted by apnicsolutions, 06-30-2008, 12:24 PM
Yes, this load is indeed high. Look at your full top process list, are any processes regularly maxing out the CPU? What hardware is inside your server?
Posted by Woooo, 06-30-2008, 12:26 PM
its dual opteron 2212 sata 250 Gb 4 GB RAM Will having half of data placed on a secondary drive help with IO?
Posted by apnicsolutions, 06-30-2008, 12:28 PM
Well I'm not sure what you're running on it, but the hardware sounds fine. Having half of your data on a secondary drive could help, but only if two conditions are met: - The secondary drive is not on the same controller as the first, otherwise it will probably just impede performance further. - The soaring load is down to drive access speeds which is unlikely. What are you running on this server?
Posted by Woooo, 06-30-2008, 12:31 PM
This server hosts my personal websites, including forums, social networks, directories etc, and a dozen client websites just staic negligible websites. What do u eamn by this: - The soaring load is down to drive access speeds which is unlikely. Are there multiple controllers for harddrives on motherboards?
Posted by The Universes, 06-30-2008, 12:32 PM
I would figure out what is causing this high iowait first, because just putting in another drive won't solve the problem for sure.
Posted by Woooo, 06-30-2008, 12:42 PM
Well this is current Top scene, iostat dont work on my machine, how can i know further details?
Posted by The Universes, 06-30-2008, 12:44 PM
Heres a problem: Tasks: 924 total Thats wayyyy to many processes running. I'm going to guess apache childs ps aux | grep httpd | wc -l iostat is a part of systat utils
Posted by Woooo, 06-30-2008, 12:47 PM
root@sl ~]# ps aux | grep httpd | wc -l 170 Yes its a busy server, and runs various busy websites, But how can i lessen the loads and less wait time? will more HDD resources help? as RAM and CPU are always available?
Posted by The Universes, 06-30-2008, 12:54 PM
Thats only 170/9xx processes. What else is running a lot of processes? Could there be someone who is abusing server resources or are you the only person with access to the server? Thats not that many apache child procs. Heres stats for a busy server I manage: Thats a lot of child procs, though the system has been tweaked and handles that no problem. Last edited by The Universes; 06-30-2008 at 12:57 PM.
Posted by Woooo, 06-30-2008, 12:59 PM
No there s just me who controlls this server in any ways. its 170 because the total processess are 600 something and some other kind of process also runs: [root@sl ~]# ps aux | grep wget | wc -l 120 [root@sl ~]# ps aux | grep mysqld | wc -l 37
Posted by The Universes, 06-30-2008, 01:02 PM
What are you wget'ing so much of? If your downloading large files, that means disk writes, which could clog up the disk.
Posted by Woooo, 06-30-2008, 01:08 PM
No its not large files, just bits of information, like 20k files. But its always running, so what would be a better option to lessen io wait?
Posted by The Universes, 06-30-2008, 01:13 PM
If you want a quick fix, you can try putting in another hard drive on a separate channel. I don't recommend it though, because its better on the long run to figure out what the problem is. However, thats only 300+/9xx processes. You really need to have someone look at the server and determine the problem. I can only make some educated guesses from here...
Posted by Woooo, 06-30-2008, 01:17 PM
Exactly, I will see if transferring half of the work on other harddrive makes some difference or not. I will take care for different channel/controller. Thanks for all the help.
Posted by The Universes, 06-30-2008, 01:21 PM
No problem. Just dont come back and blame me if it doesnt work ;-)

Add to Favourites Print this Article