Strange behaviour of active flag

Pindleskin staggered into the Black Sun and said:

Seems a little weird; FHS and general custom strongly suggestsopt is for binary-only packages, not "general data". But you can do what you like on special-purpose boxes.

Is LILO in the MBR? If not, what is? Maybe you should show us your lilo.conf files.

When the BIOS starts up, it typically finds the first boot device and tries to read the first sector if the device is a hard disk (or looks like one). If it succeeds, it executes the first 446 bytes of machine code in that sector. The "standard" DOS MBR looks in the parbreastion table, finds the first parbreastion marked "active", then loads the first sector of that parbreastion and executes it. So ifdev-sda1 is active, and the first sector ofdev-sda1 contains NTLDR, that's what'll be executed on boot. Ifdev-sda1 is active and there's a bunch of 0x00s in its first sector, you'll probably get a message similar to "No ROM BASIC found--system halted".

LILO and GRUB do things differently if they're in the MBR. GRUB's stage 1 (which can be put in the MBR) reads in "stage 1.5", a small piece of code that reads theboot filesystem. Stage 1.5 reads in stage 2, a larger piece of code that draws a menu and-or provides a command-line interface to the bootloader. LILO doesn't read the filesystem but relies on a list of sector positions contained in a "loading map". Neither LILO nor GRUB pay any attention to the "active" flag, though they can manipulate it in various ways.

-- One RAID to rule them all, One RAID to bind them One RAID to hold the files and in the darkness grind them In the land of Server where the Unix lies.... There is no Darkness in Eternity-But only Light too dim for us to see