4

On two different servers (with Ubuntu 12.04LTS AMD64) I have seen the following behaviour:

op - 10:50:05 up 305 days, 21:17,  1 user,  load average: 1.94, 2.52, 2.97
Tasks: 141 total,   2 running, 139 sleeping,   0 stopped,   0 zombie
Cpu(s): 41.5%us,  6.5%sy,  0.0%ni, 51.8%id,  0.0%wa,  0.2%hi,  0.1%si,  0.0%st
Mem:   8178432k total,  5753740k used,  2424692k free,   159480k buffers
Swap: 15625208k total,        0k used, 15625208k free,  4905292k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND                                                                               
    1 root      20   0 23928 2072 1216 S    0  0.0   0:56.42 init                                                                                  
    2 root      20   0     0    0    0 S    0  0.0   0:00.01 kthreadd                                                                              
    3 root      RT   0     0    0    0 S    0  0.0   0:01.23 migration/0                                                                           
    4 root      20   0     0    0    0 S    0  0.0   2:39.82 ksoftirqd/0                                                                           
    5 root      RT   0     0    0    0 S    0  0.0   0:00.00 watchdog/0                                                                            
    6 root      RT   0     0    0    0 S    0  0.0   0:02.99 migration/1                                                                           
    7 root      20   0     0    0    0 S    0  0.0   2:32.15 ksoftirqd/1                                                                           
    8 root      RT   0     0    0    0 S    0  0.0   0:00.00 watchdog/1                                                                            
    9 root      RT   0     0    0    0 S    0  0.0   0:11.67 migration/2                                                                           
   10 root      20   0     0    0    0 S    0  0.0  29:00.34 ksoftirqd/2

The server is working fine, but top shows all processes as using 0% CPU. A reboot fixed this on an earlier machine, but I haven't yet tried it on this one.

I have tried top several times, and so am sure that I haven't accidentally pressed '<' or '>' to sort by a different column. Sorting the process list by all of the available columns, stills shows 0% CPU for all displayed processes.

What is going on? If this a kernel bug?

Update: If I use top -p <PID> for a known, busy process, top still displays 0% CPU for that process.

Update2: My point is that ALL processes are reporting 0% CPU usage ALL of the time.

fadedbee
  • 1,988
  • 4
  • 22
  • 33
  • Have you try htop? Install and you will see what htop is showing. Maybe its some kind of bug with top – B14D3 Nov 12 '12 at 11:14
  • when you write top, push `shift + m` This will sort by memory usage. A great command ive used in the past [cpu and memory usage](http://www.commandlinefu.com/commands/view/3922/cpu-and-memory-usage-top-10-under-linux). – Sc0rian Nov 12 '12 at 11:52
  • iostat is your friend, it will show you if you have a high iowait status. If you are running debian or ubuntu, install the sysstat package. – jishi Nov 12 '12 at 12:30
  • Also, your orering seems to be according to PID, not CPU. Cpu(s): 41.5%us indicates that you have 45% CPU utilization – jishi Nov 12 '12 at 12:31
  • @jishi I'm assuming that this is because every process is using 0% CPU. – fadedbee Nov 12 '12 at 14:40
  • @chrisdew Yeah, you are probably right, however Cpu(s): 41.5%us indicates otherwise... – jishi Nov 12 '12 at 16:18
  • 0% is an approximation. Multiple values of 0.0something will add-up. – Overmind Jun 21 '19 at 08:21
  • This posting describes the same thing. Possibly the high I/O % is causing all CPU processes to stall (?) https://unix.stackexchange.com/questions/620804/kworker-flush-99-i-o-hangs-file-write-on-machine – Danny Nov 21 '20 at 10:35

2 Answers2

3

Have a look at this article. http://blog.scoutapp.com/articles/2009/07/31/understanding-load-averages

Typically Load with little to no CPU usage indicates I/O to disk/network. Load isn't a bad thing, but keeping an eye on the trends of your 1,5, and 15 min metrics will help you triage a real issue vs. trends.

Maybe check out what your disks are doing via 'iostat'.

xeux
  • 39
  • 2
2

Load isn't CPU usage. Load is "amount of runnable processes". Seeing a load of almost 2 with no CPU usage means that some processes are probably doing a lot of IO, or maybe even stuck. Check with PS whether you have processes in D state for instance.

(I had a mailserver with load 2200 last week, its storage failed. Everything else worked normally though :))

Dennis Kaarsemaker
  • 18,793
  • 2
  • 43
  • 69