As is common in filesystems, a ZFS pool's metadata and data livesin essentially a tree; at the top of the tree is the ZFS uberblockand the actual root metaobject set (which is pointed to by the ZFSuberblock). Because ZFS is a copy on write filesystem, none of thismetadata is overwritten in place. Instead, all metadata is writtento a new location, all the way up to the uberblock. This is simplefor everything except the uberblock; you write the new version of the

There are at least three different general ways that you can lose a ZFSpool.

The straightforward way that everyone knows about is for you to lose atop level vdev, ie to lose a non-redundant disk, or all of the disks ina mirror set, or enough disks in a raidzN (two disks for raidz1, threefor a raidz2, etc). Losing a chunk of striped or concatenated storage isessentially instant death for basically any RAID system and ZFS is noexception here.

(I

In yesterday's entry on how to lose ZFS pools , Iwrote:

The third way is to have corrupted metadata at the top of thepool. There are a number of ways that this can happen, but probablythe most common one is running into a ZFS bug that causes it to writeincorrect or bad data to disk [...]

I'm pretty sure I'm wrong about that and that there is a much morecommon way to get corrupt metadata: disk systems that lie to you