9-stable: one-device ZFS fails

[Arno J. Klaassen]

hello,

[ old thread omitted; kept subject ]

bon, I still see this behaviour, quite easy to reprodruce using
an external disk connected via USB (I applied r232358 to rule
out (as much as possible) other reasons.

I know this is not the dreamt setup for ZFS (Dual Pentium T3400
with 3G of main memory), I permitt myself to repost, thinking it
might indicate a problem (latest 9-stable, stripped GENERIC, initialy
nothing in /boot/loader.conf, now
   
  vfs.zfs.arc_max="983040000"
  vfs.zfs.txg.timeout="5"
  vfs.zfs.write_limit_override="536870912"

but that doesn't seem to bother).

For info :

  [root at cc ~]# camcontrol devlist | fgrep pass2
  <WDC WD50 00AAKS-00TMA0 0000>      at scbus4 target 0 lun 0 (da0,pass2)
  [root at cc ~]# dmesg -a | fgrep pass2
  pass2 at umass-sim0 bus 0 scbus4 target 0 lun 0
  pass2: <WDC WD50 00AAKS-00TMA0 0000> Fixed Direct Access SCSI-2 device 
  pass2: Serial Number 00A123456813
  pass2: 40.000MB/s transfers

Then I do :

  zpool create zdisk /dev/da0
  zfs zdisk/test
  zfs create zdisk/test
  cd /zdisk/test/
  [put some data on it]

Quite soon I get again :

  root at cc ~]# zpool status -v zdisk
    pool: zdisk
   state: ONLINE
  status: One or more devices has experienced an error resulting in data
          corruption.  Applications may be affected.
  action: Restore the file in question if possible.  Otherwise restore the
          entire pool from backup.
     see: http://www.sun.com/msg/ZFS-8000-8A
    scan: resilvered 182K in 0h0m with 0 errors on Fri Mar  2 23:02:01 2012
  config:
  
        NAME        STATE     READ WRITE CKSUM
        zdisk       ONLINE       0     1     0
          da0       ONLINE       0     1     0

  errors: Permanent errors have been detected in the following files:

        /zdisk/test/<XXX>
  [root at cc ~]# 

Same kind of error as for my original post, though this time even the
system doesn't seem to believe it :

  [root at cc ~]# cat /zdisk/test/<XXX> > /dev/null
  [root at cc ~]# md5 /zdisk/test/<XXX<
  MD5 (/zdisk/test/<XXX> = 764a10604b8c0076ec8955a2a35c0b57

( the "orginal" ... : )
  [root at cc ~]#  md5 /zgeli/home/<XXX>
  MD5 (/zgeli/home/<XXX> = 764a10604b8c0076ec8955a2a35c0b57


I made a 12-part raidz on the original disk I reported this problem on :

  [root at cc /]# gpart show         
  =>       63  976773105  ada0  MBR  (465G)
           63   79691409     1  ntfs  (38G)
     79691472   67108608     2  freebsd  [active]  (32G)
    146800080  829973088     3  freebsd  (395G)

  =>       0  67108608  ada0s2  BSD  (32G)
           0   4194304       1  freebsd-ufs  (2.0G)
     4194304   6340608       2  freebsd-swap  (3.0G)
    10534912  56573696       5  freebsd-ufs  (27G)

  =>        0  103746635  ada0s3.eli  BSD  (395G)
            0    8645542           1  freebsd-zfs  (33G)
      8645542    8645542           2  freebsd-zfs  (33G)
     17291084    8645542           4  freebsd-zfs  (33G)
     25936626    8645542           5  freebsd-zfs  (33G)
     34582168    8645542           6  freebsd-zfs  (33G)
     43227710    8645542           7  freebsd-zfs  (33G)
     51873252    8645542           8  freebsd-zfs  (33G)
     60518794    8645542           9  freebsd-zfs  (33G)
     69164336    8645542          10  freebsd-zfs  (33G)
     77809878    8645542          11  freebsd-zfs  (33G)
     86455420    8645542          12  freebsd-zfs  (33G)
     95100962    8645542          13  freebsd-zfs  (33G)
    103746504        131              - free -  (524k)

  [root at cc /]# 

Works like a charm .... :

  [root at cc /]# zpool status zgeli
    pool: zgeli
   state: ONLINE
    scan: scrub repaired 0 in 4h56m with 0 errors on Sun Feb 26 18:54:11
    2012
  config:

          NAME             STATE     READ WRITE CKSUM
          zgeli            ONLINE       0     0     0
            raidz1-0       ONLINE       0     0     0
              ada0s3.elia  ONLINE       0     0     0
              ada0s3.elib  ONLINE       0     0     0
              ada0s3.elid  ONLINE       0     0     0
              ada0s3.elie  ONLINE       0     0     0
              ada0s3.elif  ONLINE       0     0     0
              ada0s3.elig  ONLINE       0     0     0
              ada0s3.elih  ONLINE       0     0     0
              ada0s3.elii  ONLINE       0     0     0
              ada0s3.elij  ONLINE       0     0     0
              ada0s3.elik  ONLINE       0     0     0
              ada0s3.elil  ONLINE       0     0     0
              ada0s3.elim  ONLINE       0     0     0

  errors: No known data errors


I know, bis; nobody else reported this kind of a problem; but I'm really
not aware of doing something other than usual (rather than testing ZFS
on a might be underpowered notebook) ...

Willing to help,

best,

Arno