Bad Hammer zone statistics

Antony T Curtis atcurtis at gmail.com
Fri Jun 3 14:50:12 PDT 2016


Would it be fixable by using mirror-copy to a different volume and then
mirror-copy back after newfs?

On Fri, Jun 3, 2016, 14:24 Tomohiro Kusumi <kusumi.tomohiro at gmail.com>
wrote:

> Using a 5TB disk, I filled up the first layer1 (layer#0 for the first
> 4TB of the filesystem) to see if the statistics gets broken by a bug
> of statistics itself.
> As a result, it's not broken, so I guess your raid issue did break
> your filesystem metadata.
>
>
> [root@]~# df -T /HAMMER
> Filesystem  Type   1K-blocks        Used     Avail Capacity  Mounted on
> TEST        hammer 4881580032 4310634400 570945632    88%    /HAMMER
> [root@]~# umount /HAMMER
> [root@]~# hammer -vvf /dev/da4 blockmap > out
> [root@]~# grep "layer1 " out -A 5
>   layer1 4000000000000000 @2000000000800000 blocks-free 0
>         4000000000000000 zone=4  vol=0   L1#=0      L2#=0      L1=0
>    L2=0       app=8388608 free=0       fill=100.0
> crc=9d65da9d-6663e9c8
>         4000000000800000 zone=4  vol=0   L1#=0      L2#=1      L1=0
>    L2=16      app=8388608 free=0       fill=100.0
> crc=9d65da9d-6663e9c8
>         4000000001000000 zone=4  vol=0   L1#=0      L2#=2      L1=0
>    L2=32      app=8388608 free=0       fill=100.0
> crc=9d65da9d-6663e9c8
>         4000000001800000 zone=3  vol=0   L1#=0      L2#=3      L1=0
>    L2=48      app=8388608 free=0       fill=100.0
> crc=9d65da9d-12fb0047
>         4000000002000000 zone=3  vol=0   L1#=0      L2#=4      L1=0
>    L2=64      app=8388608 free=0       fill=100.0
> crc=9d65da9d-12fb0047
> --
>   layer1 4000040000000000 @2000000001000000 blocks-free 69741
>         4000040000000000 zone=10 vol=0   L1#=1      L2#=0      L1=32
>    L2=0       app=8388608 free=0       fill=100.0
> crc=5cccfffc-8f523ad6
>         4000040000800000 zone=10 vol=0   L1#=1      L2#=1      L1=32
>    L2=16      app=8388608 free=0       fill=100.0
> crc=5cccfffc-8f523ad6
>         4000040001000000 zone=10 vol=0   L1#=1      L2#=2      L1=32
>    L2=32      app=8388608 free=0       fill=100.0
> crc=5cccfffc-8f523ad6
>         4000040001800000 zone=10 vol=0   L1#=1      L2#=3      L1=32
>    L2=48      app=8388608 free=0       fill=100.0
> crc=5cccfffc-8f523ad6
>         4000040002000000 zone=10 vol=0   L1#=1      L2#=4      L1=32
>    L2=64      app=8388608 free=0       fill=100.0
> crc=5cccfffc-8f523ad6
> [root@]~# tail -22 ./out
> HAMMER zone statistics
>         zone #  blocks       items              used[B]             used[%]
>         zone 0  69741        0                  0                   0
>         zone 1  0            0                  0                   0
>         zone 2  0            0                  0                   0
>         zone 3  128          0                  1073741824          100
>         zone 4  3            0                  25165824            100
>         zone 5  0            0                  0                   0
>         zone 6  0            0                  0                   0
>         zone 7  0            0                  0                   0
>         zone 8  1075         0                  8795586560          97.5363
>         zone 9  2            0                  5610928             33.4437
>         zone 10 525078       0                  4404626128896       99.9989
>         zone 11 0            0                  0                   0
>         zone 12 0            0                  0                   0
>         zone 13 0            0                  0                   0
>         zone 14 0            0                  0                   0
>         zone 15 452549       0                  3796256161792       100
>
> ----------------------------------------------------------------------
>         total   1048576      0                  8210782395824       93.3458
> 0 bad layer1
> 0 bad layer2
>
>
> 2016-06-03 3:27 GMT+09:00 Tomohiro Kusumi <kusumi.tomohiro at gmail.com>:
> > Your total block count matches your disk size (12TB),
> > so it's either layer2's bytes_free field is broken or this statistics
> has a bug.
> >
> >
> >> M block=200004b1f4800000 zone=11 calc 5386240 free, got 5470496
> >>
> >> BM block=200005e72e000000 zone=9  calc 5999696 free, got 5999824
> >>
> >> BM block=200006400c800000 zone=10 calc 7995392 free, got 8001360
> >>
> >> BM block=2000064225000000 zone=10 calc 8192000 free, got 8257536
> >>
> >> BM block=2000064225800000 zone=10 calc 8192000 free, got 8126464
> >>
> >> BM block=200006445d800000 zone=10 calc 8126464 free, got 8060928
> >>
> >> BM block=200006445f000000 zone=10 calc 8323072 free, got 8257536
> >>
> >> BM block=2000077674000000 zone=10 calc 6864896 free, got 7192576
> >
> > These errors come from the reason I mentioned above.
> >
> > Technically speaking, these offsets says the bad data are stored in
> > the second layer1 entry, where 1 layer1 entry is 4TB.
> > In other words these are beyond 4TB of the filesystem.
> > I'll check if hammer blockmap/checkmap have a bug in this case
> > tomorrow, before I say your metadata is broken.
> >
> >
> > 2016-06-03 2:54 GMT+09:00 Antony T Curtis <atcurtis at gmail.com>:
> >> Overall size of the disk is around 12 Tb.
> >>
> >> slightly different result from checkmap, likely because it is a mounted
> >> volume...
> >>
> >>> M block=200004b1f4800000 zone=11 calc 5386240 free, got 5470496
> >>>
> >>> BM block=200005e72e000000 zone=9  calc 5999696 free, got 5999824
> >>>
> >>> BM block=200006400c800000 zone=10 calc 7995392 free, got 8001360
> >>>
> >>> BM block=2000064225000000 zone=10 calc 8192000 free, got 8257536
> >>>
> >>> BM block=2000064225800000 zone=10 calc 8192000 free, got 8126464
> >>>
> >>> BM block=200006445d800000 zone=10 calc 8126464 free, got 8060928
> >>>
> >>> BM block=200006445f000000 zone=10 calc 8323072 free, got 8257536
> >>>
> >>> BM block=2000077674000000 zone=10 calc 6864896 free, got 7192576
> >>>
> >>> HAMMER zone statistics
> >>>
> >>> zone #  blocks       items              used[B]             used[%]
> >>>
> >>> zone 0  0            0                  0                   0
> >>>
> >>> zone 1  0            0                  0                   0
> >>>
> >>> zone 2  0            0                  0                   0
> >>>
> >>> zone 3  128          0                  1073741824          100
> >>>
> >>> zone 4  4            0                  33554432            100
> >>>
> >>> zone 5  0            0                  0                   0
> >>>
> >>> zone 6  0            0                  0                   0
> >>>
> >>> zone 7  0            0                  0                   0
> >>>
> >>> zone 8  1029         0                  7988363264          92.5449
> >>>
> >>> zone 9  1443         0                  2183889824          18.0416
> >>>
> >>> zone 10 330736       0                  2911421278384       104.938
> >>>
> >>> zone 11 1154         0                  14288724672         147.604
> >>>
> >>> zone 12 0            0                  0                   0
> >>>
> >>> zone 13 0            0                  0                   0
> >>>
> >>> zone 14 0            0                  0                   0
> >>>
> >>> zone 15 0            0                  0                   0
> >>>
> >>> ----------------------------------------------------------------------
> >>>
> >>> total   334494       0                  2936989552400       104.67
> >>>
> >>> 0 bad nodes
> >>>
> >>> 8 errors
> >>
> >>
> >>
> >> On 2 June 2016 at 10:42, Tomohiro Kusumi <kusumi.tomohiro at gmail.com>
> wrote:
> >>>
> >>> 1. What's your approximate disk size (or sum of disks) ?
> >>>
> >>> 2. If your total block counts from blockmap output matches your disk
> size,
> >>> > zone 0  1095221      0                  18446744073709486080
> 2.00784e+06
> >>> this line from blockmap output shows your layer2 metadata probably had
> >>> bad (broken) free_bytes.
> >>> Can't tell how it was broken from this output though.
> >>>
> >>> 3. What are the 10 errors for checkmap ?
> >>> You have output before statistics that indicates those errors.
> >>>
> >>>
> >>> 2016-06-03 2:32 GMT+09:00 Antony T Curtis <atcurtis at gmail.com>:
> >>> > Hammer checkmap shows this instead:
> >>> >>
> >>> >> HAMMER zone statistics
> >>> >>
> >>> >> zone #  blocks       items              used[B]             used[%]
> >>> >>
> >>> >> zone 0  0            0                  0                   0
> >>> >>
> >>> >> zone 1  0            0                  0                   0
> >>> >>
> >>> >> zone 2  0            0                  0                   0
> >>> >>
> >>> >> zone 3  128          0                  1073741824          100
> >>> >>
> >>> >> zone 4  4            0                  33554432            100
> >>> >>
> >>> >> zone 5  0            0                  0                   0
> >>> >>
> >>> >> zone 6  0            0                  0                   0
> >>> >>
> >>> >> zone 7  0            0                  0                   0
> >>> >>
> >>> >> zone 8  1029         0                  7988092928          92.5418
> >>> >>
> >>> >> zone 9  1443         0                  2183767648          18.0406
> >>> >>
> >>> >> zone 10 330734       0                  2911406172336       104.938
> >>> >>
> >>> >> zone 11 1154         0                  14287486768         147.591
> >>> >>
> >>> >> zone 12 0            0                  0                   0
> >>> >>
> >>> >> zone 13 0            0                  0                   0
> >>> >>
> >>> >> zone 14 0            0                  0                   0
> >>> >>
> >>> >> zone 15 0            0                  0                   0
> >>> >>
> >>> >>
> ----------------------------------------------------------------------
> >>> >>
> >>> >> total   334492       0                  2936972815936       104.67
> >>> >>
> >>> >> 0 bad nodes
> >>> >>
> >>> >> 10 errors
> >>> >
> >>> >
> >>> >
> >>> > On 2 June 2016 at 10:14, Antony T Curtis <atcurtis at gmail.com> wrote:
> >>> >>
> >>> >> This is from a hammer blockmap command.
> >>> >>
> >>> >> On 2 June 2016 at 10:13, Tomohiro Kusumi <kusumi.tomohiro at gmail.com
> >
> >>> >> wrote:
> >>> >>>
> >>> >>> From which command is this from ?
> >>> >>> It's not hammer show (because hammer show has non 0 items), so I
> >>> >>> assume either blockmap or checkmap.
> >>> >>>
> >>> >>> 2016-06-03 2:05 GMT+09:00 Antony T Curtis <atcurtis at gmail.com>:
> >>> >>> > Is there any info as to how to repair a Hammer volume or should I
> >>> >>> > simply try
> >>> >>> > to backup and restore?
> >>> >>> >
> >>> >>> >> HAMMER zone statistics
> >>> >>> >>
> >>> >>> >> zone #  blocks       items              used[B]
>  used[%]
> >>> >>> >>
> >>> >>> >> zone 0  1095221      0                  18446744073709486080
> >>> >>> >> 2.00784e+06
> >>> >>> >>
> >>> >>> >> zone 1  0            0                  0                   0
> >>> >>> >>
> >>> >>> >> zone 2  0            0                  0                   0
> >>> >>> >>
> >>> >>> >> zone 3  128          0                  1073741824          100
> >>> >>> >>
> >>> >>> >> zone 4  4            0                  33554432            100
> >>> >>> >>
> >>> >>> >> zone 5  0            0                  0                   0
> >>> >>> >>
> >>> >>> >> zone 6  0            0                  0                   0
> >>> >>> >>
> >>> >>> >> zone 7  0            0                  0                   0
> >>> >>> >>
> >>> >>> >> zone 8  1029         0                  7986941952
> 92.5284
> >>> >>> >>
> >>> >>> >> zone 9  1443         0                  2183756816
> 18.0405
> >>> >>> >>
> >>> >>> >> zone 10 330680       0                  2910941826784
>  104.939
> >>> >>> >>
> >>> >>> >> zone 11 1154         0                  14287415824
>  147.59
> >>> >>> >>
> >>> >>> >> zone 12 0            0                  0                   0
> >>> >>> >>
> >>> >>> >> zone 13 0            0                  0                   0
> >>> >>> >>
> >>> >>> >> zone 14 0            0                  0                   0
> >>> >>> >>
> >>> >>> >> zone 15 143205       0                  1201290608640       100
> >>> >>> >>
> >>> >>> >>
> >>> >>> >>
> ----------------------------------------------------------------------
> >>> >>> >>
> >>> >>> >> total   1572864      0                  4137797780736
>  31.3609
> >>> >>> >>
> >>> >>> >> 0 bad layer1
> >>> >>> >>
> >>> >>> >> 0 bad layer2
> >>> >>> >
> >>> >>> >
> >>> >>> >
> >>> >>> >
> >>> >>> > --
> >>> >>> > Antony T Curtis
> >>> >>> >
> >>> >>
> >>> >>
> >>> >>
> >>> >>
> >>> >> --
> >>> >> Antony T Curtis
> >>> >> 0523 C487 9187 6972 6894
> >>> >> AEC7 3087 F819 B477 B687
> >>> >
> >>> >
> >>> >
> >>> >
> >>> > --
> >>> > Antony T Curtis
> >>> > 0523 C487 9187 6972 6894
> >>> > AEC7 3087 F819 B477 B687
> >>
> >>
> >>
> >>
> >> --
> >> Antony T Curtis
> >> 0523 C487 9187 6972 6894
> >> AEC7 3087 F819 B477 B687
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.dragonflybsd.org/pipermail/users/attachments/20160603/005e9cc0/attachment-0002.html>


More information about the Users mailing list