ZFS: i/o error - all block copies unavailable after upgrading to r225312

[Henri Hennebert]

On 09/05/2011 19:51, Sebastian Chmielewski wrote:
> On Fri, 02 Sep 2011 17:01:56 +0200
> Henri Hennebert<hlh at restart.be>  wrote:
>
>> I encounter the same problem. I reboot with mfsBSD and run zpool scrub
>> without error.
>>
>> I reboot and get the same error - the error msg repeating 4 or 5 times.
>>
>
> Do you have lzjb compression in your pool

Yes on some file systems but not on rpool/root

> My pool does use it (on some filesystems) and it's also almost full (95% used).

I think it is not a problem, zfsloader must open the zfs file system in 
read-only. BUT it is AWFUL for io performance.

> 'zdb zroot' fails with 'assertion failed' message.
>
> 1089791    1    16K    64K    64K    64K  100.00  ZFS plain file
>
>
> Traversing all blocks to verify checksums and verify nothing leaked ...
> Assertion failed: (space_map_load(&msp->ms_map,&zdb_space_map_ops, SM_ALLOC,&msp->ms_smo, spa->spa_meta_objset) == 0), file /work/opensource/OS/FreeBSD/current-gem/cddl/usr.sbin/zdb/../../../cddl/contrib/opensolaris/cmd/zdb/zdb.c, line 2115.
> zsh: abort      zdb zroot

I try it on my active pool and no assertion error, only some

zdb_blkptr_cb: Got error 122 reading <180, 0, 0, 0>  -- skipping

but on a live file system this seems normal.

Henri
>
> best regards,