MrToddsFriends
Documentation Browser
- Joined
- Jan 12, 2015
- Messages
- 1,338
This is usually caused by people not understanding the two tables that GPT has and treating it like the old MBR format, which only had one table.
Others have suggested using the FreeNAS GUI. I agree with that, it does several things that are well-tested and create vdevs and pools that are well-formed and in the format that FreeNAS expects. In this case, now would be a good time to fix these problems, before a lot of data has been copied to the pool.
As mentioned earlier in this thread I saw the same GEOM messages during replacement of the two devices making up my boot pool.
To be more precise: I replaced two USB sticks (ZFS mirror) by two SSDs (ZFS mirror) using the procedure outlined in section "8.1.11. Replacing Drives to Grow a ZFS Pool" of the manual, using the GUI for all steps. I'm pretty sure that I also had a open root shell via SSH, but only for diagnostic purposes (and taking care for expansion via
zpool online -e
after replacement).So there might be a problem in the implementation of the disk replacement procedure in FreeNAS, at least in 11.0-U2, which I used at that time.
Beyond the already described behavior (showing the aforementioned GEOM messages) the replacement procedure implicated an additional problem on my system:
After the replacement the SSDs used as targets make up the current freenas-boot pool but are not bootable. When trying to boot from one of them the well known "This is a FreeNAS data disk and can not boot system. System halted." message is shown. Currently I'm booting the system using the USB-Sticks (previous freenas-boot pool), probably utilizing one or more GRUB stages installed on them AND the SSDs (current freenas-boot pool), using the boot environments installed there.
What would be the steps to make the SSDs used as replacement targets bootable belatedly? How can I find out what's missing on the SSDs exactly? I know, reinstalling would solve this, too.