Re: Is there a known fio_gettime issue?

I should note that this is running on a few drives in parallel (all in one
conf file, the same stanza repeated for each drive). I also tried with the
verify stuff turned off and got the same results. Some drives report 0 for
time and bandwidth but each drive reports the correct amount of data
transferred.

On Nov 19, 2007 11:26 PM, Shawn Lewis <shawnlewis_at_google.com> wrote:

>
>
> On Nov 19, 2007 8:29 PM, Shawn Lewis <shawnlewis_at_google.com> wrote:
>
> > Hi,
> >
> > Is there a known issue with fio_gettime or other timing code? I'm seeing
> > times of either 0 or way off for long running phases like the one listed
> > below. Before I start debugging fio (this could have something to do with
> > the verify code I added that's in the main tree) I just wanted to ask if
> > there was a known issue because I saw the debugging code in fio_gettime as
> > well as the "timewarp bug on some kernels?" comment in time.c.
> >
> > [hdc-sscan_write]
> > filename=/export/hdw3/datafile.tmp
> > rw=write
> > bs=64k
> > size=300g
> > ioengine=sync
> > bwavgtime=5000
> > direct=1
> > thread=1
> > do_verify=0
> > verify=meta
> > verify_pattern=0x55aaa55a
> > verify_interval=512
> >
> > Thanks,
> > Shawn
> >
>
>
Received on Tue Nov 20 2007 - 08:28:15 CET