I have an older DL360 Gen8 which has a notorious raid card forbidding HBA/IT mode, or rather you cannot boot if enabled. The default install of Harvester, even the installation disk has heavy I/O. I started to do chroot/grub, but honestly had not messed with image/immutable OS boots before and could not get the chroot properly done. So.. I installed Harvester once to the USB drive itself. Then installed to just the nvme drive, by itself. Which is not bootable by the server. After booting up(Coming into the USB installation), I changed the labels of COS_STATE and COST_STATE_PERSISTENT, and HARV_LH_DEFAULT on the USB drive, so that the nvme partitions would be found instead, containing the correct labels. This now boots properly, and seeming 0 I/O on the USB stick. Curious if this is clever or setting myself up for a massive failure during an upgrade -- thinking harvester uses some of the cfg files from the initial to know where to expect things when upgrading?

I am not really sure i understand the sequence of steps, please correct me if i am wrong: • installed harvester to usb drive • re-installed harvester to nvme (not sure what this means?) • change labels on usb so grub finds the labels off nvme and uses those for boot?