Mike,

Even though your disk is 13,529,147,392 bytes we will align this on a 
256K boundary. This should make the end of the disk 13,528,989,696 (you 
can verify this by looking at vdev_psize) making the offset of label 1 
at 13,528,465,408. So the offset you're seeing is an 8k offset from 
label 1.

Hope this helps,
George

Michael Noeth wrote:
> Hello,
>
> When I am importing a pool I'm seeing some weird behavior.  During the import 
> process, a call to vdev_file_probe_io is made to ensure the member vdevs are 
> functioning correctly.  From my understanding, this should occur within 
> "safe" places within the vdev label.  
>
> Given my disk layout as such:
> Disk size =  13,529,147,392 bytes
> vdev labels 0 & 2: 0 bytes - 524288 bytes
> vedv labels 1 & 3: 13,528,623,104 - 13,529,147,392
>
> When I trace through, I am seeing vdev_file_probe_io called with:
>
> vdev_file_probe_io(nvd, vl_boot, VDEV_BOOT_HEADER_SIZE, offset, UIO_WRITE);
>
> where offset = 13,528,473,600
>
> This is outside the vdev label in the data area of the disk.  I'm still 
> pretty new to ZFS and wasn't sure if the test writes should be constrained to 
> the vdev labels.
>
> Thanks,
> ~Mike
>   


Reply via email to