* [f2fs-dev] Discard issue
@ 2020-05-26 1:59 Jaegeuk Kim
2020-05-26 2:20 ` Chao Yu
0 siblings, 1 reply; 7+ messages in thread
From: Jaegeuk Kim @ 2020-05-26 1:59 UTC (permalink / raw)
To: yuchao0; +Cc: Linux F2FS Dev Mailing List
Hi Chao,
I'm hitting segment.c:1065 when running longer fsstress (1000s) with error
injection. Do you have any issue from your side?
Thanks,
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [f2fs-dev] Discard issue
2020-05-26 1:59 [f2fs-dev] Discard issue Jaegeuk Kim
@ 2020-05-26 2:20 ` Chao Yu
2020-05-26 2:26 ` Jaegeuk Kim
0 siblings, 1 reply; 7+ messages in thread
From: Chao Yu @ 2020-05-26 2:20 UTC (permalink / raw)
To: Jaegeuk Kim; +Cc: Linux F2FS Dev Mailing List
Hi Jaegeuk,
On 2020/5/26 9:59, Jaegeuk Kim wrote:
> Hi Chao,
>
> I'm hitting segment.c:1065 when running longer fsstress (1000s) with error
(1000s) do you mean time in single round or total time of multi rounds?
> injection. Do you have any issue from your side?
I haven't hit that before, in my test, in single round, fsstress won't last long
time (normally about 10s+ for each round).
Below is por_fsstress() implementation in my code base:
por_fsstress()
{
_fs_opts
while true; do
ltp/fsstress -x "echo 3 > /proc/sys/vm/drop_caches" -X 10 -r -f fsync=8 -f sync=0 -f write=4 -f dwrite=2 -f truncate=6 -f allocsp=0 -f bulkstat=0 -f bulkstat1=0 -f freesp=0 -f zero=1 -f collapse=1 -f insert=1 -f resvsp=0 -f unresvsp=0 -S t -p 20 -n 200000 -d $TESTDIR/test &
sleep 10
src/godown $TESTDIR
killall fsstress
sleep 5
umount $TESTDIR
if [ $? -ne 0 ]; then
for i in `seq 1 50`
do
umount $TESTDIR
if [ $? -eq 0]; then
break
fi
sleep 5
done
fi
echo 3 > /proc/sys/vm/drop_caches
_fsck
_mount f2fs
rm $TESTDIR/testfile
touch $TESTDIR/testfile
umount $TESTDIR
_fsck
_mount f2fs
_rm_50
done
}
Did you update this code?
Could you share more test configuration, like mkfs option, device size, mount option,
new por_fsstress() implementation if it exists? I can try to reproduce this issue
in my env.
Thanks,
>
> Thanks,
> .
>
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [f2fs-dev] Discard issue
2020-05-26 2:20 ` Chao Yu
@ 2020-05-26 2:26 ` Jaegeuk Kim
2020-05-26 7:44 ` Chao Yu
0 siblings, 1 reply; 7+ messages in thread
From: Jaegeuk Kim @ 2020-05-26 2:26 UTC (permalink / raw)
To: Chao Yu; +Cc: Linux F2FS Dev Mailing List
On 05/26, Chao Yu wrote:
> Hi Jaegeuk,
>
> On 2020/5/26 9:59, Jaegeuk Kim wrote:
> > Hi Chao,
> >
> > I'm hitting segment.c:1065 when running longer fsstress (1000s) with error
>
> (1000s) do you mean time in single round or total time of multi rounds?
>
> > injection. Do you have any issue from your side?
>
> I haven't hit that before, in my test, in single round, fsstress won't last long
> time (normally about 10s+ for each round).
>
> Below is por_fsstress() implementation in my code base:
>
> por_fsstress()
> {
> _fs_opts
>
> while true; do
> ltp/fsstress -x "echo 3 > /proc/sys/vm/drop_caches" -X 10 -r -f fsync=8 -f sync=0 -f write=4 -f dwrite=2 -f truncate=6 -f allocsp=0 -f bulkstat=0 -f bulkstat1=0 -f freesp=0 -f zero=1 -f collapse=1 -f insert=1 -f resvsp=0 -f unresvsp=0 -S t -p 20 -n 200000 -d $TESTDIR/test &
> sleep 10
> src/godown $TESTDIR
> killall fsstress
> sleep 5
> umount $TESTDIR
> if [ $? -ne 0 ]; then
> for i in `seq 1 50`
> do
> umount $TESTDIR
> if [ $? -eq 0]; then
> break
> fi
> sleep 5
> done
> fi
> echo 3 > /proc/sys/vm/drop_caches
> _fsck
> _mount f2fs
> rm $TESTDIR/testfile
> touch $TESTDIR/testfile
> umount $TESTDIR
> _fsck
> _mount f2fs
> _rm_50
> done
> }
>
> Did you update this code?
>
> Could you share more test configuration, like mkfs option, device size, mount option,
> new por_fsstress() implementation if it exists? I can try to reproduce this issue
> in my env.
I just changed, in __run_godown_fsstress(), sleep 1000 instead of 10.
https://github.com/jaegeuk/xfstests-f2fs/blob/f2fs/run.sh#L249
./run.sh por_fsstress
>
> Thanks,
>
> >
> > Thanks,
> > .
> >
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [f2fs-dev] Discard issue
2020-05-26 2:26 ` Jaegeuk Kim
@ 2020-05-26 7:44 ` Chao Yu
2020-05-27 1:44 ` Chao Yu
0 siblings, 1 reply; 7+ messages in thread
From: Chao Yu @ 2020-05-26 7:44 UTC (permalink / raw)
To: Jaegeuk Kim; +Cc: Linux F2FS Dev Mailing List
On 2020/5/26 10:26, Jaegeuk Kim wrote:
> On 05/26, Chao Yu wrote:
>> Hi Jaegeuk,
>>
>> On 2020/5/26 9:59, Jaegeuk Kim wrote:
>>> Hi Chao,
>>>
>>> I'm hitting segment.c:1065 when running longer fsstress (1000s) with error
>>
>> (1000s) do you mean time in single round or total time of multi rounds?
>>
>>> injection. Do you have any issue from your side?
>>
>> I haven't hit that before, in my test, in single round, fsstress won't last long
>> time (normally about 10s+ for each round).
>>
>> Below is por_fsstress() implementation in my code base:
>>
>> por_fsstress()
>> {
>> _fs_opts
>>
>> while true; do
>> ltp/fsstress -x "echo 3 > /proc/sys/vm/drop_caches" -X 10 -r -f fsync=8 -f sync=0 -f write=4 -f dwrite=2 -f truncate=6 -f allocsp=0 -f bulkstat=0 -f bulkstat1=0 -f freesp=0 -f zero=1 -f collapse=1 -f insert=1 -f resvsp=0 -f unresvsp=0 -S t -p 20 -n 200000 -d $TESTDIR/test &
>> sleep 10
>> src/godown $TESTDIR
>> killall fsstress
>> sleep 5
>> umount $TESTDIR
>> if [ $? -ne 0 ]; then
>> for i in `seq 1 50`
>> do
>> umount $TESTDIR
>> if [ $? -eq 0]; then
>> break
>> fi
>> sleep 5
>> done
>> fi
>> echo 3 > /proc/sys/vm/drop_caches
>> _fsck
>> _mount f2fs
>> rm $TESTDIR/testfile
>> touch $TESTDIR/testfile
>> umount $TESTDIR
>> _fsck
>> _mount f2fs
>> _rm_50
>> done
>> }
>>
>> Did you update this code?
>>
>> Could you share more test configuration, like mkfs option, device size, mount option,
>> new por_fsstress() implementation if it exists? I can try to reproduce this issue
>> in my env.
>
> I just changed, in __run_godown_fsstress(), sleep 1000 instead of 10.
>
> https://github.com/jaegeuk/xfstests-f2fs/blob/f2fs/run.sh#L249
>
> ./run.sh por_fsstress
Reproducing...
Thanks,
>
>>
>> Thanks,
>>
>>>
>>> Thanks,
>>> .
>>>
> .
>
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [f2fs-dev] Discard issue
2020-05-26 7:44 ` Chao Yu
@ 2020-05-27 1:44 ` Chao Yu
2020-05-27 1:58 ` Jaegeuk Kim
0 siblings, 1 reply; 7+ messages in thread
From: Chao Yu @ 2020-05-27 1:44 UTC (permalink / raw)
To: Jaegeuk Kim; +Cc: Linux F2FS Dev Mailing List
On 2020/5/26 15:44, Chao Yu wrote:
> On 2020/5/26 10:26, Jaegeuk Kim wrote:
>> On 05/26, Chao Yu wrote:
>>> Hi Jaegeuk,
>>>
>>> On 2020/5/26 9:59, Jaegeuk Kim wrote:
>>>> Hi Chao,
>>>>
>>>> I'm hitting segment.c:1065 when running longer fsstress (1000s) with error
>>>
>>> (1000s) do you mean time in single round or total time of multi rounds?
>>>
>>>> injection. Do you have any issue from your side?
>>>
>>> I haven't hit that before, in my test, in single round, fsstress won't last long
>>> time (normally about 10s+ for each round).
>>>
>>> Below is por_fsstress() implementation in my code base:
>>>
>>> por_fsstress()
>>> {
>>> _fs_opts
>>>
>>> while true; do
>>> ltp/fsstress -x "echo 3 > /proc/sys/vm/drop_caches" -X 10 -r -f fsync=8 -f sync=0 -f write=4 -f dwrite=2 -f truncate=6 -f allocsp=0 -f bulkstat=0 -f bulkstat1=0 -f freesp=0 -f zero=1 -f collapse=1 -f insert=1 -f resvsp=0 -f unresvsp=0 -S t -p 20 -n 200000 -d $TESTDIR/test &
>>> sleep 10
>>> src/godown $TESTDIR
>>> killall fsstress
>>> sleep 5
>>> umount $TESTDIR
>>> if [ $? -ne 0 ]; then
>>> for i in `seq 1 50`
>>> do
>>> umount $TESTDIR
>>> if [ $? -eq 0]; then
>>> break
>>> fi
>>> sleep 5
>>> done
>>> fi
>>> echo 3 > /proc/sys/vm/drop_caches
>>> _fsck
>>> _mount f2fs
>>> rm $TESTDIR/testfile
>>> touch $TESTDIR/testfile
>>> umount $TESTDIR
>>> _fsck
>>> _mount f2fs
>>> _rm_50
>>> done
>>> }
>>>
>>> Did you update this code?
>>>
>>> Could you share more test configuration, like mkfs option, device size, mount option,
>>> new por_fsstress() implementation if it exists? I can try to reproduce this issue
>>> in my env.
>>
>> I just changed, in __run_godown_fsstress(), sleep 1000 instead of 10.
>>
>> https://github.com/jaegeuk/xfstests-f2fs/blob/f2fs/run.sh#L249
>>
>> ./run.sh por_fsstress
>
> Reproducing...
After one night reproducing, the issue still not occur..
BTW, I enabled below features in image:
extra_attr project_quota inode_checksum flexible_inline_xattr inode_crtime compression
and tagged compression flag on root inode.
>
> Thanks,
>
>>
>>>
>>> Thanks,
>>>
>>>>
>>>> Thanks,
>>>> .
>>>>
>> .
>>
>
>
> _______________________________________________
> Linux-f2fs-devel mailing list
> Linux-f2fs-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
> .
>
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [f2fs-dev] Discard issue
2020-05-27 1:44 ` Chao Yu
@ 2020-05-27 1:58 ` Jaegeuk Kim
2020-05-27 2:06 ` Chao Yu
0 siblings, 1 reply; 7+ messages in thread
From: Jaegeuk Kim @ 2020-05-27 1:58 UTC (permalink / raw)
To: Chao Yu; +Cc: Linux F2FS Dev Mailing List
On 05/27, Chao Yu wrote:
> On 2020/5/26 15:44, Chao Yu wrote:
> > On 2020/5/26 10:26, Jaegeuk Kim wrote:
> >> On 05/26, Chao Yu wrote:
> >>> Hi Jaegeuk,
> >>>
> >>> On 2020/5/26 9:59, Jaegeuk Kim wrote:
> >>>> Hi Chao,
> >>>>
> >>>> I'm hitting segment.c:1065 when running longer fsstress (1000s) with error
> >>>
> >>> (1000s) do you mean time in single round or total time of multi rounds?
> >>>
> >>>> injection. Do you have any issue from your side?
> >>>
> >>> I haven't hit that before, in my test, in single round, fsstress won't last long
> >>> time (normally about 10s+ for each round).
> >>>
> >>> Below is por_fsstress() implementation in my code base:
> >>>
> >>> por_fsstress()
> >>> {
> >>> _fs_opts
> >>>
> >>> while true; do
> >>> ltp/fsstress -x "echo 3 > /proc/sys/vm/drop_caches" -X 10 -r -f fsync=8 -f sync=0 -f write=4 -f dwrite=2 -f truncate=6 -f allocsp=0 -f bulkstat=0 -f bulkstat1=0 -f freesp=0 -f zero=1 -f collapse=1 -f insert=1 -f resvsp=0 -f unresvsp=0 -S t -p 20 -n 200000 -d $TESTDIR/test &
> >>> sleep 10
> >>> src/godown $TESTDIR
> >>> killall fsstress
> >>> sleep 5
> >>> umount $TESTDIR
> >>> if [ $? -ne 0 ]; then
> >>> for i in `seq 1 50`
> >>> do
> >>> umount $TESTDIR
> >>> if [ $? -eq 0]; then
> >>> break
> >>> fi
> >>> sleep 5
> >>> done
> >>> fi
> >>> echo 3 > /proc/sys/vm/drop_caches
> >>> _fsck
> >>> _mount f2fs
> >>> rm $TESTDIR/testfile
> >>> touch $TESTDIR/testfile
> >>> umount $TESTDIR
> >>> _fsck
> >>> _mount f2fs
> >>> _rm_50
> >>> done
> >>> }
> >>>
> >>> Did you update this code?
> >>>
> >>> Could you share more test configuration, like mkfs option, device size, mount option,
> >>> new por_fsstress() implementation if it exists? I can try to reproduce this issue
> >>> in my env.
> >>
> >> I just changed, in __run_godown_fsstress(), sleep 1000 instead of 10.
> >>
> >> https://github.com/jaegeuk/xfstests-f2fs/blob/f2fs/run.sh#L249
> >>
> >> ./run.sh por_fsstress
> >
> > Reproducing...
>
> After one night reproducing, the issue still not occur..
>
> BTW, I enabled below features in image:
>
> extra_attr project_quota inode_checksum flexible_inline_xattr inode_crtime compression
>
> and tagged compression flag on root inode.
Could you check disk supports discard? I didn't set compression to the root
inode.
I set _mkfs with "f2fs":
mkfs.f2fs -f -O encrypt -O extra_attr -O quota -O inode_checksum /dev/$DEV;;
# run.sh reload
# run.sh por_fsstress
>
> >
> > Thanks,
> >
> >>
> >>>
> >>> Thanks,
> >>>
> >>>>
> >>>> Thanks,
> >>>> .
> >>>>
> >> .
> >>
> >
> >
> > _______________________________________________
> > Linux-f2fs-devel mailing list
> > Linux-f2fs-devel@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
> > .
> >
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [f2fs-dev] Discard issue
2020-05-27 1:58 ` Jaegeuk Kim
@ 2020-05-27 2:06 ` Chao Yu
0 siblings, 0 replies; 7+ messages in thread
From: Chao Yu @ 2020-05-27 2:06 UTC (permalink / raw)
To: Jaegeuk Kim; +Cc: Linux F2FS Dev Mailing List
On 2020/5/27 9:58, Jaegeuk Kim wrote:
> On 05/27, Chao Yu wrote:
>> On 2020/5/26 15:44, Chao Yu wrote:
>>> On 2020/5/26 10:26, Jaegeuk Kim wrote:
>>>> On 05/26, Chao Yu wrote:
>>>>> Hi Jaegeuk,
>>>>>
>>>>> On 2020/5/26 9:59, Jaegeuk Kim wrote:
>>>>>> Hi Chao,
>>>>>>
>>>>>> I'm hitting segment.c:1065 when running longer fsstress (1000s) with error
>>>>>
>>>>> (1000s) do you mean time in single round or total time of multi rounds?
>>>>>
>>>>>> injection. Do you have any issue from your side?
>>>>>
>>>>> I haven't hit that before, in my test, in single round, fsstress won't last long
>>>>> time (normally about 10s+ for each round).
>>>>>
>>>>> Below is por_fsstress() implementation in my code base:
>>>>>
>>>>> por_fsstress()
>>>>> {
>>>>> _fs_opts
>>>>>
>>>>> while true; do
>>>>> ltp/fsstress -x "echo 3 > /proc/sys/vm/drop_caches" -X 10 -r -f fsync=8 -f sync=0 -f write=4 -f dwrite=2 -f truncate=6 -f allocsp=0 -f bulkstat=0 -f bulkstat1=0 -f freesp=0 -f zero=1 -f collapse=1 -f insert=1 -f resvsp=0 -f unresvsp=0 -S t -p 20 -n 200000 -d $TESTDIR/test &
>>>>> sleep 10
>>>>> src/godown $TESTDIR
>>>>> killall fsstress
>>>>> sleep 5
>>>>> umount $TESTDIR
>>>>> if [ $? -ne 0 ]; then
>>>>> for i in `seq 1 50`
>>>>> do
>>>>> umount $TESTDIR
>>>>> if [ $? -eq 0]; then
>>>>> break
>>>>> fi
>>>>> sleep 5
>>>>> done
>>>>> fi
>>>>> echo 3 > /proc/sys/vm/drop_caches
>>>>> _fsck
>>>>> _mount f2fs
>>>>> rm $TESTDIR/testfile
>>>>> touch $TESTDIR/testfile
>>>>> umount $TESTDIR
>>>>> _fsck
>>>>> _mount f2fs
>>>>> _rm_50
>>>>> done
>>>>> }
>>>>>
>>>>> Did you update this code?
>>>>>
>>>>> Could you share more test configuration, like mkfs option, device size, mount option,
>>>>> new por_fsstress() implementation if it exists? I can try to reproduce this issue
>>>>> in my env.
>>>>
>>>> I just changed, in __run_godown_fsstress(), sleep 1000 instead of 10.
>>>>
>>>> https://github.com/jaegeuk/xfstests-f2fs/blob/f2fs/run.sh#L249
>>>>
>>>> ./run.sh por_fsstress
>>>
>>> Reproducing...
>>
>> After one night reproducing, the issue still not occur..
>>
>> BTW, I enabled below features in image:
>>
>> extra_attr project_quota inode_checksum flexible_inline_xattr inode_crtime compression
>>
>> and tagged compression flag on root inode.
>
> Could you check disk supports discard? I didn't set compression to the root
> inode.
I start to review discard support codes from yesterday, however I have not found
anything suspectable yet.
>
> I set _mkfs with "f2fs":
> mkfs.f2fs -f -O encrypt -O extra_attr -O quota -O inode_checksum /dev/$DEV;;
Let me update test configs.
Thanks,
>
> # run.sh reload
> # run.sh por_fsstress
>
>>
>>>
>>> Thanks,
>>>
>>>>
>>>>>
>>>>> Thanks,
>>>>>
>>>>>>
>>>>>> Thanks,
>>>>>> .
>>>>>>
>>>> .
>>>>
>>>
>>>
>>> _______________________________________________
>>> Linux-f2fs-devel mailing list
>>> Linux-f2fs-devel@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
>>> .
>>>
> .
>
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2020-05-27 2:10 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-05-26 1:59 [f2fs-dev] Discard issue Jaegeuk Kim
2020-05-26 2:20 ` Chao Yu
2020-05-26 2:26 ` Jaegeuk Kim
2020-05-26 7:44 ` Chao Yu
2020-05-27 1:44 ` Chao Yu
2020-05-27 1:58 ` Jaegeuk Kim
2020-05-27 2:06 ` Chao Yu
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).