delaywrite test failures

[Jeremy Allison]

On Fri, Feb 07, 2014 at 09:24:21AM +0100, Stefan (metze) Metzmacher wrote:
> Am 07.02.2014 00:40, schrieb Jeremy Allison:
> > where immediately is set as:
> > 
> >         int used_delay = torture_setting_int(tctx, "writetimeupdatedelay", 2000000);
> >         int normal_delay = 2000000;
> >         double sec = ((double)used_delay) / ((double)normal_delay);
> > 
> > (where sec ends up as 0.25 probably due to the writetimeupdatedelay setting
> > in the test scripts)
> > 
> >         if (diff > (0.25 * sec * 0.75)) { /* 0.75 to cope with vmware timing */
> > 			ERROR....
> > 
> > So I'm guessing this needs relaxing some for the test
> > environment - looks like our 'immediately' is too strict
> > in the test harness. (0.25 * 0.25 * 0.75 == 0.0625),
> 
> 0.25 * 0.25 = 0.0625

Thanks, missed the last multiplication :-).

> 0.25 * 0.25 * 0.75 = 0,046875
> 
> I'm wondering why * 0.75 because of vmware timing,
> shouldn't that be + 0.75?

That might be the correct fix...

> (0.25 * sec) + (0.75 * sec) should be much better
> 
> with sec = 0.25, it would be 0,25
> with sec = 1, it would be 1.
> 
> There're 2 cases:
> 
> - we need to make sure the server updates without delay,
>   (we should check that the responses comes after <= half of
>    the writetimeupdatedelay time)
> - or we need to make sure response is received after >= writetimeupdatedelay
> 
> Then we might want to add hack for crazy clocks.

I'll take a look at this once I've finished my code reviews.

Jeremy.