How reliable is top?

pid()

CPU states:  cpu    user    nice  system    irq  softirq  iowait    idle
           total   12.0%    0.0%    6.4%   0.0%     0.8%    0.0%  779.2%
           cpu00    0.4%    0.0%    0.0%   0.0%     0.0%    0.0%   99.5%
           cpu01    3.9%    0.0%    0.9%   0.0%     0.4%    0.0%   94.6%
           cpu02    0.0%    0.0%    0.0%   0.0%     0.0%    0.0%  100.0%
           cpu03    0.0%    0.0%    0.0%   0.0%     0.0%    0.0%  100.0%
           cpu04    0.9%    0.0%    0.4%   0.0%     0.0%    0.0%   98.5%
           cpu05    0.4%    0.0%    0.0%   0.0%     0.0%    0.0%   99.5%
           cpu06    4.4%    0.0%    4.9%   0.0%     0.0%    0.0%   90.6%
           cpu07    1.9%    0.0%    0.4%   0.0%     0.4%    0.0%   97.0%

Notice the 779.2% total idle. I have a sister box - which is an exact replica of this machine and it's as idle as this one, yet it's total is 98.5% idle.

djmarkus

i prefer htop

pid()

A very imformative post djmarkus.

Why?

Khannie

What OS is this on?

For linux, you can get details from /proc relatively easily. Top uses libgtop iirc, and is considered pretty solid. There is a solaris port of it, but of course, solaris collects resource usage information differently to linux.

htop almost certainly uses the same library, but it provides more information.

pid()

Khannie wrote: »

What OS is this on?

For linux, you can get details from /proc relatively easily. Top uses libgtop iirc, and is considered pretty solid. There is a solaris port of it, but of course, solaris collects resource usage information differently to linux.

htop almost certainly uses the same library, but it provides more information.

Thanks Khannie. It's Redhat Enterprise Server, running 2.4.21-50.ELsmp.

Khannie

I'm guessing that's RHEL3. I think that's the last one that used a 2.4 kernel. That 779 looks like a refresh issue to me. Pure speculation, but I'd say it's 79.2% idle or thereabouts.

pid()

Yep, RHEL3. I'm convinced it's not a refresh issue. Notice what it says for total user % here:

           total  112.8%    0.0%   56.8%   0.0%    24.0%    0.0%  604.0%
           cpu00    7.3%    0.0%    5.3%   0.0%     0.0%    0.0%   87.2%
           cpu01   19.7%    0.0%    5.4%   0.0%     5.4%    0.0%   69.4%
           cpu02    9.8%    0.0%    6.8%   0.0%     2.9%    0.0%   80.3%
           cpu03    6.3%    0.0%    5.8%   0.0%     1.4%    0.0%   86.2%
           cpu04   22.5%    0.0%    8.8%   0.0%     2.9%    0.0%   65.6%
           cpu05   22.0%    0.0%   13.7%   0.0%     4.9%    0.0%   59.3%
           cpu06   14.2%    0.0%    8.3%   0.0%     1.4%    0.0%   75.9%
           cpu07   11.2%    0.0%    2.9%   0.4%     4.9%    0.0%   80.3%

It's at 112.8%. It was at 300% a while ago. Hehe, I'm stumped.

EDIT: Also note this from the Redhat people:

Redhat wrote:

An average percent of CPU time. This field examines all aspects of CPU time, including user, idle, and niced tasks. Because niced tasks are included, the total can go over 100%, so do not be alarmed if this is the case (unless it is an outrageous value, such as 160%).

Looks like it's some sort of issue.

Dorsanty

No one else think 'total' = 'sum of the values' for the different CPUs? A total is numerically different to an average. So it's not lying to you.


So if you are 100% idle on all CPUs and have 8 CPUs you are 800% idle, i.e. you seriously need to consider why you've payed for an 8 CPU system that's not doing anything.

pid()

Dorsanty wrote: »

No one else think 'total' = 'sum of the values' for the different CPUs? A total is numerically different to an average. So it's not lying to you.


So if you are 100% idle on all CPUs and have 8 CPUs you are 800% idle, i.e. you seriously need to consider why you've payed for an 8 CPU system that's not doing anything.

And the winner is.... Dorsanty!

bman

pid() wrote: »

And the winner is.... Dorsanty!

I'm not so sure. Here's the output of top on a machine we have at work with 4 CPUs:

CPU states:  cpu    user    nice  system    irq  softirq  iowait    idle
           total    5.3%    0.0%    1.6%   0.0%     0.9%   21.5%   70.4%
           cpu00    5.3%    0.0%    0.7%   0.0%     3.1%   24.5%   66.1%
           cpu01    8.9%    0.0%    1.9%   0.0%     0.1%   23.3%   65.5%
           cpu02    4.7%    0.0%    1.9%   0.0%     0.1%    0.0%   93.0%            
           cpu03    2.3%    0.0%    1.9%   0.0%     0.1%   38.2%   57.2%

total here is the average of all 4 CPUs.

Khannie

Hmmmm, which version of top is that lads? I'm running "procps version 3.2.7" at home (gentoo) and I only see 1 CPU state for my quad core. It says

Cpu(s):  0.0%us,  0.0%sy,  0.0%ni,100.0%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st

edit: Found it. You press 1 to toggle. Me like.

pid()

bman wrote: »

I'm not so sure. Here's the output of top on a machine we have at work with 4 CPUs:

CPU states:  cpu    user    nice  system    irq  softirq  iowait    idle
           total    5.3%    0.0%    1.6%   0.0%     0.9%   21.5%   70.4%
           cpu00    5.3%    0.0%    0.7%   0.0%     3.1%   24.5%   66.1%
           cpu01    8.9%    0.0%    1.9%   0.0%     0.1%   23.3%   65.5%
           cpu02    4.7%    0.0%    1.9%   0.0%     0.1%    0.0%   93.0%            
           cpu03    2.3%    0.0%    1.9%   0.0%     0.1%   38.2%   57.2%

total here is the average of all 4 CPUs.

Yeah, I dunno... Here's the output from the two sister machines, which are identical:

Machine 1:

           CPU states:  cpu    user    nice  system    irq  softirq  iowait    idle
           total    3.2%    0.0%    1.6%   0.0%     0.0%    0.0%  793.6%
           cpu00    0.4%    0.0%    0.0%   0.0%     0.0%    0.0%   99.5%
           cpu01    0.9%    0.0%    0.0%   0.0%     0.0%    0.0%   99.0%
           cpu02    0.0%    0.0%    0.0%   0.0%     0.0%    0.0%  100.0%
           cpu03    0.0%    0.0%    0.9%   0.0%     0.0%    0.0%   99.0%
           cpu04    0.4%    0.0%    0.4%   0.0%     0.0%    0.0%   99.0%
           cpu05    0.0%    0.0%    0.0%   0.0%     0.0%    0.0%  100.0%
           cpu06    0.4%    0.0%    0.0%   0.0%     0.0%    0.0%   99.5%
           cpu07    1.4%    0.0%    0.4%   0.0%     0.4%    0.0%   97.5%

Machine 2:

           CPU states:  cpu    user    nice  system    irq  softirq  iowait    idle
           total    0.1%    0.0%    0.1%   0.0%     0.1%    0.0%   99.3%
           cpu00    0.5%    0.0%    0.3%   0.0%     0.0%    0.0%   99.0%
           cpu01    0.3%    0.0%    0.7%   0.0%     0.5%    0.0%   98.2%
           cpu02    0.1%    0.0%    0.0%   0.0%     0.0%    0.0%   99.8%
           cpu03    0.0%    0.0%    0.0%   0.0%     0.0%    0.0%  100.0%
           cpu04    0.0%    0.0%    0.0%   0.0%     0.0%    0.0%  100.0%
           cpu05    0.0%    0.0%    0.0%   0.0%     0.0%    0.0%  100.0%
           cpu06    0.3%    0.0%    0.0%   0.0%     0.0%    0.0%   99.6%
           cpu07    0.0%    0.0%    0.0%   0.5%     0.9%    0.0%   98.4%

But essentially, if you add up the idle percentages on machine 1, you get the total percentage at the top. Perhaps someone has just changed the way top is viewing this on machine 1.

bman

I suppose unless you use top for some important reason you'll manage with it totalling all the CPUs percentages. Probably something in your configuration rather than a bug.