-2

I need to fix my virtual server. What's wrong with it? Needs memory? or else?

4 core with 4GB memory

top - 10:06:13 up 53 min,  1 user,  load average: 31.74, 30.17, 24.79
Tasks: 180 total,   1 running, 179 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.9%us,  0.2%sy,  0.0%ni, 69.3%id, 29.6%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   4194304k total,  2753136k used,  1441168k free,        0k buffers
Swap:  4194304k total,        0k used,  4194304k free,  1060564k cached
  • What does your server run? – Faheem Mitha Apr 7 '16 at 3:32
  • CentOS Linux 6.5. Many online store there. – Screenager8 Apr 7 '16 at 3:50
  • No, sorry. I meant: what kind of jobs does your server do? E.g. web server? – Faheem Mitha Apr 7 '16 at 3:54
  • Yes... web server – Screenager8 Apr 7 '16 at 3:57
1

From the information you provided I can only tell your server needs a faster disk.

You could try to use atop to see which program does the most disk IO. If you are lucky you can configure that program to do less IO.

| improve this answer | |
  • An alternative to atop is iotop. – Faheem Mitha Apr 7 '16 at 7:42
0

For a discussion on how load averages are calculated, look at the somewhat dated (but still useful) article --

 http://www.linuxjournal.com/article/9001?page=0,1

From your data, since your load avg ~ 24 - 32 and you have 4 cores, the average load / core is ~ 6 - 8 (processes waiting for CPU). Since your CPU utilization is low (69% idle), we can only conclude that you have a large no of threads waiting for I/O to complete (the load avg calculation includes threads waiting for I/O to complete). The high wait % (~30) also seems to indicate the same. There are 2 things you could do - 1. run iostat on the disks to see detailed stats on per disk wait times. 2. run ps and check for 'D' in that status to see threads waiting for I/O to complete. The next step to solving the above (depending on where the problem is) could range from better and faster disks or taking a closer look at the I/O paths in your code. You could use other tools such as perf to gain even deeper insights into where exactly the application wait times come from.

| improve this answer | |

Not the answer you're looking for? Browse other questions tagged or ask your own question.