I’m testing out some planned changes in a VirtualBox VM before I apply them to my real server. My main aim is to replace my boot drives (2x NVME in mirror) with other drives (SATA SSDs) in my system. I’ve therefore mocked up the following:
- Scale 24.04
- 2x20GB in mirror as boot pool
- 2x 200GB as spare SSDs
- 2x 1TB in mirror as HDD pool
I moved the system dataset from the HDDs to the boot pool, then rebooted OK.
I then replaced one of the 20GB drives in the boot pool with a 200GB drive and rebooted. It took a very long time to shut down, seemingly because middlewared was unresponsive. I don’t think this is relevant to the problem, though.
I then replaced the second 20GB drive with the remaining 200GB drive and rebooted again. This reboot took a normal amount of time and the system came back up as normal.
Finally, I created a new pool with the leftover 20GB drives: I pointed the apps service at this new pool, replicated the app’s config folder across and updated the app to use it and confirmed that it was running correctly; then I deleted the now redundant app config dataset and the old ix-applications dataset.
I immediately rebooted, and VirtualBox reported that there were no bootable media. I got around this by accessing the VM’s bios and picking other HDDs until I found one (actually two) that worked.
My assumption is that when TrueNAS is initially installed, it marks its boot media as bootable, but when I used TrueNAS to replace those marked drives, it didn’t copy/transfer that bootable flag to the replacements. Is this likely to be correct? If so, then is there something I can do in TrueNAS, before, during or after the replacement, to mark the new disks as bootable?
Alternatively, if this happens with my physical server, I guess I’ll need to change the boot order in the BIOS or maybe edit the bootloader’s priority, assuming I can work out where to do it.
Thanks in advance.