I just recently upgraded the coolers on my Supermicro server and now it can't find the default boot device unless I unplug ALL of the drives except the boot drive.
This system is an older Supermicro SC846, 24-drive bay, 4U machine. I took it out of the rack, replaced the stock CPU heatsinks with some larger ones that have fans, then re-assembled and re-inserted the drives (in the same order they came out, they're labeled).
Now when I turn it on, the system boots to a screen:
Reboot and Select proper Boot device or insert boot media in selected Boot device and press a key
All of the drives are in the same bay as they always have been, but now the RAID controller won't load it by default.
The system has an LSI 3ware 9650SE-24M8. There are 12 hard drives and 1 SSD. The SSD is (and always has been) in slot 0 of the array and on the controller.
I went through the BIOS, but there's nothing that indicates which drive is the boot drive. Some internet searches indicates that the drive in Slot 0 should be the boot volume and I've confirmed this configuration.
The OS is Debian 9 with grub installed on
/dev/sda, which correlates to slot-0 on the controller. I've also tried reinstalling Grub from Debian Rescue, but that had no effect. I also ran grub-update once I got into the system, but that didn't help either; I'm confident this is more of a BIOS issue, but I don't know how to fix it if so.
I found the manual for the 3ware controller which contains this information:
If you will install your OS on a disk or unit attached to the 3ware RAID controller, specify the controller as the boot device. (Note that if you configured more than one unit, the drive(s) specified as Unit 0 will be treated as the boot disk.)
Position the unit you want to be bootable at the top of the list of exportable units.
The boot disk is in slot 0. Is there something I need to do to refresh the boot-flag on the volume? Would this help with the BIOS when it scans the disks?
I have no idea how to fix this.