Re: RFC on io-stalls patch

[Jens Axboe]

On Tue, Jul 15 2003, Chris Mason wrote:
> On Tue, 2003-07-15 at 04:28, Jens Axboe wrote:
> 
> > Definitely, because prepare to be a bit disappointed. Here are scores
> > that include 2.4.21 as well:
> 
> > io_load:
> > Kernel            [runs]        Time    CPU%    Loads   LCPU%   Ratio
> > 2.4.21                 3        543     49.7    100.4   19.0    4.08
> > 2.4.22-pre5            3        637     42.5    120.2   18.5    4.75
> > 2.4.22-pre5-axboe      3        540     50.0    103.0   18.1    4.06
> 
> Huh, this is completely different than io_load on my box (2P scsi, ext3,
> data=writeback)
> 
> io_load:
> Kernel      [runs]      Time    CPU%    Loads   LCPU%   Ratio
> 2.4.21           3      520     52.5    27.8    15.2    3.80
> 2.4.22-pre5      3      394     69.0    21.5    15.4    2.90
> 2.4.22-sync      3      321     84.7    16.2    15.8    2.36
> 
> Where 2.4.22-sync was the variant I posted yesterday.  I don't really
> see how 2.4.21 can get numbers as good as 2.4.22-pre5 on the io_load
> test, the read starvation with a big streaming io is horrible.
> 
> The data=writeback is changing the workload significantly, I used it
> because I didn't want the data=ordered code to flush all dirty buffers
> every 5 seconds.  I would expect ext3 data=ordered to be pretty
> starvation prone in 2.4.21 as well though.

Well maybe it's due to data=writeback? I'm using completely stock
options for ext3. You didn't mention tcq depth of your scsi drive, in my
experience it's worthless to test io scheduler behaviour using more than
a few tags.

> BTW, the contest run times vary pretty wildy.  My 3 compiles with
> io_load running on 2.4.21 were 603s, 443s and 515s.  This doesn't make
> the average of the 3 numbers invalid, but we need a more stable metric.

Mine are pretty consistent [1], I'd suspect that it isn't contest but your
drive tcq skewing things. But it would be nice to test with other things
as well, I just used contest because it was at hand.

[1] when the current run completes, I can post them all.

-- 
Jens Axboe