M. Edward (Ed) Borasky wrote:
> I'm experimenting with "btt" and I have a case where the "iostat"
> emulation is generating negative queue lengths. They seem to be steadily
> increasing in magnitude. The utilizations don't look right, either -- I
> don't think this device was 100 percent busy for the entire run.
> 
> I've attached a sample -- the input files are huge, but I'll post them
> if you need them.
> 
> 

Hi Ed -

If you get a chance, could you try the attached patch - I have not
committed it, although it looks solid enough. I'm going to keep plugging
on this. [Note the stuff about await & svctm in the patch description.]

Thanks,
Alan