On Wed, Jul 29, 2009 at 12:07 PM, Scott Carey <[email protected]>wrote:

>
> On 7/29/09 11:47 AM, "Todd Lipcon" <[email protected]> wrote:
>
> > On Wed, Jul 29, 2009 at 2:17 AM, Steve Loughran <[email protected]>
> wrote:
> >
> >> Scott Carey wrote:
> >>
> >>> Well, the first thing to do in any performance bottleneck investigation
> is
> >>> to look at the machine hardware resource usage.
> >>>
> >>> During your test, what is the CPU use and disk usage?  What about
> network
> >>> utilization?
> >>> Top, vmstat, iostat, and some network usage monitoring would be useful.
> >>>  It
> >>> could be many things causing your lack of scalability, but without
> >>> actually
> >>> monitoring your machines to see if there is an obvious bottleneck its
> just
> >>> random guessing and hunches.
> >>>
> >>
> >> iotop is very good here
> >>
> >
> > But sadly not available with kernels before 2.6.20 :( I was looking into
> > this recently and the next best thing I could find was blktrace, but it's
> > not nearly as nice.
> >
> > Any other tools people know about, especially that work on RHEL5/CentOS5?
> >
>
> Iftop worked for me for network traffic:
> http://www.ex-parrot.com/pdw/iftop/
>
> For disk monitoring on CentOS/RHEL 5 Linux, every time I try and drill down
> a problem with iostat I wish I was on Windows or Solaris instead where
> monitoring and tracing disk I/O isn't a decade behind.  On both of those,
> not only can you see what process is causing the disk activity, you can
> find
> out what FILES the activity is on.
>

Yep, dtrace is great for this. I imagine if one were clever they could use
strace to look at IO patterns on a per-process level. If anyone knows of any
handy wrappers to do this, that would be great for looking at performance
issues like this.

-Todd

Reply via email to