Re: computing load averages

From: Ben Russo (ben_at_muppethouse.com)
Date: 05/19/05

  • Next message: Ben Russo: "Re: computing load averages" 
    Date: Thu, 19 May 2005 13:23:57 -0400
To: General Red Hat Linux discussion list <redhat-list@redhat.com>

    Shragai, Yaron wrote:
    > Hello,
    > How are load averages computed for the top command and the /proc/loadavg
    > file? Where do the numbers come from?
    >
    > Thanks,
    > Yaron
    >

    cat /proc/loadavg

    These are numbers internal to the kernel.

    They are 5, 10 and 15 minute average values
    of the number of processes that are in a "run" state at each sample.

    You should consider your load in reference to the number of CPU's you
    have in a system. If you have 8 CPU's, then having a load of 7 isn't a
    problem. If the load is higher than the number of CPU's then you should
    look at your top's CPU stats, and look at iostat and vmstat and sar.

    A box can have a high load for several different reasons (or
    combinations of those reasons). You may have a number of processes all
    running simultaneously and the CPU is just running as fast as possible
    (evidenced by less than 2% idle time shown in top for CPU usage).

    You may have a number of processes that are in run states, but are
    waiting for data i/o (evidenced by having abundant CPU idle time, but
    still having a load higher than the number of CPU's).

    Check vmstat for memory swap activity, i/o activity

    -Ben. 

    -- 
redhat-list mailing list
unsubscribe mailto:redhat-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/redhat-list
  • Next message: Ben Russo: "Re: computing load averages"

    Relevant Pages

    • Re: computing load averages
      ... Shragai, Yaron wrote: ... You should consider your load in reference to the number of CPU's you ... look at your top's CPU stats, and look at iostat and vmstat and sar. ... Check vmstat for memory swap activity, ...
      (RedHat)
    • Re: Fwd: Re: New Metrics to measure Load average
      ... School of Electrical and Information Engineering ... Currently for the Phd, I am using division of load as per login users and this is what I want to have for HPC and for the "Task Profiling Model". ... Be very wary if their planned CPU controller as the model that they intend to use will only work with CPU bound processes. ...
      (Linux-Kernel)
    • [PATCH 2/2] sched: Improve fairness of cpu allocation for task groups
      ... The current load balancing scheme isn't good for group fairness. ... Col4 -> CPU bandwidth received by each individual task of the group. ... Currently cpu's weight is just the summation of individual task weights. ...
      (Linux-Kernel)
    • [PATCH] sched: Consolidated and improved smpnice patch
      ... The introduction of separate run queues per CPU has brought with it "nice" enforcement problems that are best described by a simple example. ... whether this expectation is met is pretty much down to luck as there are four equally likely distributions of the tasks to the CPUs that the load balancing code will consider to be balanced with loads of 2.0 for each CPU. ... To overcome this problem SCHED_LOAD_SCALE has been replaced by the load_weight for the relevant task or by the average load_weight per task for the queue in question. ...
      (Linux-Kernel)
    • [PATCH 2/2] sched: Improve fairness of cpu allocation for task groups
      ... The current load balancing scheme isn't good for group fairness. ... Col4 -> CPU bandwidth received by each individual task of the group. ... Currently cpu's weight is just the summation of individual task weights. ...
      (Linux-Kernel)