[Bug 651846] Re: raid10 fails - "lost page write due to I/O error on md4" and " raid10_make_request bug: can't convert block across chunks or bigger than 128k 1623343324 20" - write fails, remote nfs mount of filesystem becomes unusable

[Andrew Hately]

Following the above, I did this:

1) stop nfs server daemon, unmount /home, stop /dev/md4
2) make a new md4 with default chunk size (64KiB) but otherwise the same as above
3) made a new xfs filesystem on md4 using md3 as log device, as before
4) mount this as /home, export via nfs
5) restore 1.2TB of backed up data
6) copy several large files over the nfs mount
And there are no problems.

So the only significant difference: default chunk size OK, non-default chunk size fails.
I guess somewhere in the raid10 code an assumption is made about chunk size.

Some documentation of what I did:
 mdadm -C /dev/md4 -v -n 6 -l 10 -p f2 -f /dev/sd[abcdef]3
mdadm: chunk size defaults to 64K
mdadm: /dev/sda3 appears to be part of a raid array:
    level=raid10 devices=6 ctime=Tue Sep 28 14:34:19 2010
mdadm: /dev/sdb3 appears to be part of a raid array:
    level=raid10 devices=6 ctime=Tue Sep 28 14:34:19 2010
mdadm: /dev/sdc3 appears to be part of a raid array:
    level=raid10 devices=6 ctime=Tue Sep 28 14:34:19 2010
mdadm: /dev/sdd3 appears to be part of a raid array:
    level=raid10 devices=6 ctime=Tue Sep 28 14:34:19 2010
mdadm: /dev/sde3 appears to be part of a raid array:
    level=raid10 devices=6 ctime=Tue Sep 28 14:34:19 2010
mdadm: /dev/sdf3 appears to be part of a raid array:
    level=raid10 devices=6 ctime=Tue Sep 28 14:34:19 2010
mdadm: size set to 961980160K
Continue creating array? y
mdadm: array /dev/md4 started.

root at wibert:~# mkfs.xfs -f -d su=65536,sw=6 -l logdev=/dev/md3 -L home /dev/md4
meta-data=/dev/md4               isize=256    agcount=32, agsize=22546416 blks
         =                       sectsz=512   attr=2
data     =                       bsize=4096   blocks=721485120, imaxpct=5
         =                       sunit=16     swidth=96 blks
naming   =version 2              bsize=4096   ascii-ci=0
log      =/dev/md3               bsize=4096   blocks=32112, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0

root at wibert:~# mdadm --detail /dev/md4
/dev/md4:
        Version : 00.90
  Creation Time : Thu Sep 30 14:27:19 2010
     Raid Level : raid10
     Array Size : 2885940480 (2752.25 GiB 2955.20 GB)
  Used Dev Size : 961980160 (917.42 GiB 985.07 GB)
   Raid Devices : 6
  Total Devices : 7
Preferred Minor : 4
    Persistence : Superblock is persistent

    Update Time : Fri Oct  1 09:21:19 2010
          State : clean
 Active Devices : 6
Working Devices : 7
 Failed Devices : 0
  Spare Devices : 1

         Layout : near=1, far=2
     Chunk Size : 64K

           UUID : 4e420fd5:e9f8a4ff:fcc517c0:fe041f9d (local to host wibert)
         Events : 0.38

    Number   Major   Minor   RaidDevice State
       0       8        3        0      active sync   /dev/sda3
       1       8       19        1      active sync   /dev/sdb3
       2       8       35        2      active sync   /dev/sdc3
       3       8       51        3      active sync   /dev/sdd3
       4       8       67        4      active sync   /dev/sde3
       5       8       83        5      active sync   /dev/sdf3

       6       8      147        -      spare   /dev/sdj3


When did non-default chunk size last work on raid10?
http://blog.jamponi.net/2008/07/raid56-and-10-benchmarks-on-26255_10.html

-- 
raid10 fails - "lost page write due to I/O error on md4" and " raid10_make_request bug: can't convert block across chunks or bigger than 128k 1623343324 20" - write fails, remote nfs mount of filesystem becomes unusable
https://bugs.launchpad.net/bugs/651846
You received this bug notification because you are a member of Kernel
Bugs, which is subscribed to linux in ubuntu.