Not having much luck with RAID-Z boot

Thomas Backman serenity at
Thu May 28 17:27:27 UTC 2009

I'm trying out the new RAID-Z boot support, but I'm not having much  
luck. After finally getting VMware to get the boot order right (and  
not move on to PXE boot etc.) all I get is errors. See the original  

I followed in the footsteps on the first poster (maybe not such a  
great idea since it failed for him), even to the point of using three  
disks in VMware Fusion, although that was a coincidence.
Anyway, I have a healthy "rpool" made out of three GPT partitioned  
disks, as in the post. The only thing I changed was $1 to the disk  
names (da2, da3, da4) and the size parameter for the freebsd-zfs  
I then populated the pool with everything from my current r192914  
install, and made sure that zpool.cache was up to date in /rpool/boot/ 
zfs (by copying the current copy there, and making sure it contained  

All I get is "error 4 lba xxxxxxxxxxx", 30+ times, followed by the  
loader starting to load the kernel, and failing with 10-15 lines of  
"ZFS: i/o error - all block copies unavailable". It can't load the  
kernel, manually or not. It does read loader.conf successfully though  
(since vfs.root.mountfrom and vm.kmem_max etc. are set correctly).

System: FreeBSD 8.0-CURRENT (DTRACE) #4 r192914: Thu May 28 08:56:46  
CEST 2009

Built with LOADER_ZFS_SUPPORT=YES in /etc/make.conf, no patches  
applied (since the raidz boot patch was merged to current a few days  

Any advice?


More information about the freebsd-current mailing list