On Mon, Apr 29, 2024 at 08:59:54AM -0700, Darrick J. Wong wrote:
> On Sun, Apr 28, 2024 at 01:49:53AM +0800, Zorro Lang wrote:
> > On Sat, Apr 27, 2024 at 09:55:30AM +0200, Christoph Hellwig wrote:
> > > generic/095 doesn't take more than 4 seconds on any of my test setups,
> >
> > Yeah, it generally takes ~5 seconds. But there's a "LOAD_FACTOR", so
>
> On what kind of storage hardware? If it's that fast even on spinning
> rust then I guess you could double the iodepth*/numjobs computations.
On most of our testing machines, even very old (10+ years ago) with
general HP scsi disk, nothing special.
# lshw -class disk
*-disk
description: SCSI Disk
product: LOGICAL VOLUME
vendor: HP
physical id: 1.0.0
bus info: scsi@2:1.0.0
logical name: /dev/sda
version: 6.00
serial: xxxxxxxxxxxxxxx
size: 931GiB (1TB)
capabilities: 15000rpm gpt-1.00 partitioned partitioned:gpt
# ./check generic/095
FSTYP -- xfs (non-debug)
PLATFORM -- Linux/x86_64 hp-xxxx-xxx 6.9.0-0.rc5.20240425gite88c4cfcb7b8.47.fc41.x86_64 #1 SMP PREEMPT_DYNAMIC Thu Apr 25 14:21:52 UTC 2024
MKFS_OPTIONS -- -f /dev/sda4
MOUNT_OPTIONS -- -o context=system_u:object_r:root_t:s0 /dev/sda4 /mnt/scratch
generic/095 2s ... 2s
Ran: generic/095
Passed all 1 tests
>
> (What happens if numjobs exceeds the 9 defined in the fio config file?)
# export LOAD_FACTOR=3; ./check generic/095
FSTYP -- xfs (non-debug)
PLATFORM -- Linux/x86_64 hp-dl380pg8-01 6.9.0-0.rc5.20240425gite88c4cfcb7b8.47.fc41.x86_64 #1 SMP PREEMPT_DYNAMIC Thu Apr 25 14:21:52 UTC 2024
MKFS_OPTIONS -- -f /dev/sda4
MOUNT_OPTIONS -- -o context=system_u:object_r:root_t:s0 /dev/sda4 /mnt/scratch
generic/095 3s ... 3s
Ran: generic/095
Passed all 1 tests
# export LOAD_FACTOR=4; ./check generic/095
FSTYP -- xfs (non-debug)
PLATFORM -- Linux/x86_64 hp-dl380pg8-01 6.9.0-0.rc5.20240425gite88c4cfcb7b8.47.fc41.x86_64 #1 SMP PREEMPT_DYNAMIC Thu Apr 25 14:21:52 UTC 2024
MKFS_OPTIONS -- -f /dev/sda4
MOUNT_OPTIONS -- -o context=system_u:object_r:root_t:s0 /dev/sda4 /mnt/scratch
generic/095 3s ... - output mismatch (see /root/git/xfstests/results//generic/095.out.bad)
--- tests/generic/095.out 2024-03-02 20:15:52.580273536 +0800
+++ /root/git/xfstests/results//generic/095.out.bad 2024-04-30 02:46:27.874906090 +0800
@@ -1,2 +1,6 @@
QA output created by 095
+fio: first I/O failed. If /mnt/scratch/file1 is a zoned block device, consider --zonemode=zbd
+fio: first I/O failed. If /mnt/scratch/file1 is a zoned block device, consider --zonemode=zbd
+fio: first I/O failed. If /mnt/scratch/file1 is a zoned block device, consider --zonemode=zbd
+fio: first I/O failed. If /mnt/scratch/file1 is a zoned block device, consider --zonemode=zbd
Silence is golden
...
(Run 'diff -u /root/git/xfstests/tests/generic/095.out /root/git/xfstests/results//generic/095.out.bad' to see the entire diff)
Ran: generic/095
Failures: generic/095
Failed 1 of 1 tests
This failure looks like: https://lore.kernel.org/fstests/a8f03f2c-5951-4aea-856b-218326815302@xxxxxxxxx/
# rpm -q fio
fio-3.37-1.fc41.x86_64
>
> > before we add it into quick group, I'd like to ask
> > do you think if it's worth increasing the default LOAD_FACTOR a bit to
> > give it more stress, or move it to quick group directly?
>
> I'd move it to the quick group.
OK, Christoph likes that too, let's move it to quick.
Reviewed-by: Zorro Lang <zlang@xxxxxxxxxx>
Thanks,
Zorro
>
> --D
>
> > > but is exercises code that handles buffered write iterations interrupted
> > > by concurrent direct I/O that no other test in the quick group does.
> > >
> > > Signed-off-by: Christoph Hellwig <hch@xxxxxx>
> > > ---
> > > tests/generic/095 | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/tests/generic/095 b/tests/generic/095
> > > index 80448200b..9d0446e98 100755
> > > --- a/tests/generic/095
> > > +++ b/tests/generic/095
> > > @@ -7,7 +7,7 @@
> > > # Concurrent mixed I/O (buffer I/O, aiodio, mmap, splice) on the same files
> > > #
> > > . ./common/preamble
> > > -_begin_fstest auto rw stress
> > > +_begin_fstest auto quick rw stress
> > >
> > > # Import common functions.
> > > . ./common/filter
> > > --
> > > 2.39.2
> > >
> > >
> >
> >
>
