dtrace discuss - May 2006 - iotime.d - Bug ID: 6345226

I copy-pasted the following code in DTrace Guide to my own I/O script,
with which I noticed my disk response time are too good to be true,
i.e. it seems I got more than what I paid for :-)

"this->elapsed / 10000000"

This nsec to msec conversion is off by 10, turns out to be a 6-month-old bug:
http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6345226

It''s in both current DTrace Guide and /usr/demo/dtrace/iotime.d file.

Shall we leave this bug alone so that folks feel really good about I/O
performance on Solaris?
Just kidding :-)

Tao

Hi Tao,

This slipped between the cracks somehow. Next time someone puts together a
DTrace wad, we''ll be sure to include a fix for this bug.

Adam

On Fri, May 12, 2006 at 02:33:17AM -0500, Tao Chen
wrote:
> I copy-pasted the following code in DTrace Guide to my own I/O script,
> with which I noticed my disk response time are too good to be true,
> i.e. it seems I got more than what I paid for :-)
> 
> "this->elapsed / 10000000"
> 
> This nsec to msec conversion is off by 10, turns out to be a 6-month-old 
> bug:
> http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6345226
> 
> It''s in both current DTrace Guide and /usr/demo/dtrace/iotime.d
file.
> 
> Shall we leave this bug alone so that folks feel really good about I/O
> performance on Solaris?
> Just kidding :-)
> 
> Tao
> _______________________________________________
> dtrace-discuss mailing list
> dtrace-discuss at opensolaris.org
-- 
Adam Leventhal, Solaris Kernel Development       http://blogs.sun.com/ahl