Opened 14 years ago
#8974 new defect
--bootable=off doesn't apply to IDE controller
| Reported by: | Bayard Bell | Owned by: | |
|---|---|---|---|
| Component: | virtual disk | Version: | VirtualBox 4.0.8 |
| Keywords: | Cc: | ||
| Guest type: | other | Host type: | other |
Description
In deciding what's Hard Disk 0, the controller order is internal. IDE controllers are first in that order, and they can't be excluded, unlike every other type of controller. I don't know that the use case for this is tremendous, so it might be sufficient simply to document it. I was trying to use it in benchmarking to determine the performance difference between SATA and IDE, particularly because some recent bugs have had people suggesting that one ought at least try it. Solaris in particular has problems dealing with changes to the path to its boot device (grub will find the rpool and bootfs, but boot subsequently breaks because other config files store the path to the device), so things break if I've previously installed with a SATA boot VDI but want to try a benchmark with data stored on an IDE data VDI.
I already discussed the particulars on this one with MichalN on #vbox.
VBox.log for a failed boot