> > I think this goes too far, and I don't see what it would change. > IIRC, it has been mentioned that grub-install works flawlessly when > slave drive is available and even when slave drive is removed and > system is booted with bootdisk. However, the newly written GRUB then > fails in MBR as soon as slave drive is not available. > >
this was in response solely to Ashleys GRUB_INSTALLS not apparently having an effect - this is only a theory, but it seems a reasonable explanation frogive me if I am redundant here what I am saying is that the MBR may be crafted or rebuild by anaconda for the two drive setup disk info. Anaconda may not be able to go backwards here, may not be able to undo in a return to single drive configuration - most people add hardware, not remove it, it may have not been a tested situation. "Something" changed on the linux system in response to the addition of the hard drive that is not being undone when he reboots with only one drive installed. If the contents of the MBR in regards to the location of the Phase 1.5 image are the real issue here, then replacing it with a freshly rebuilt version would be required in order to correct the issue. "Grub-install". which I have not seen admittedly, may be based on a MAKE environment strategy. IF SO IT ONLY REBUILDS The MBR if and only if some local disk file has changed that a dependency has been declared for. It may not check to see if the hardware environment changed. By cleansing out the existing config info via a MAKE CLEAN or CLOBBER sequence , you should force it to do a fresh rebuild from scratch on the MBR program which should insert the new single drive related information. ie SO FAR your newly written grub is the same image as the old grub that was on the mbr to begin with. You may have changed nothing by running grub-install in regards to the MBR image on sector 0. Ashley just rewrote the same 512 byte block over and over again. To rephrase, What I am saying is the copy of the MBR on drive A is the same as the copy of the MBR image on /boot that grub-install stuffs back onto the MBR, so all the grub installs changed nothing, just rewrote the same info over and over again unchanged for the new single drive configuration he was trying to configure for.. Popping the hard drive on/off changes the underlying hardware/bios info such that it matched/"did not match" the contents of the MBR. So it worked any time he had the drive configuration matching the corresponding info in the MBR. One would need to rebuild the MBR based on the new single drive architecture to correct this issue - recompile from sources as it were after booting to a single drive setup. The boot floppy does not use the MBR, It is the MBR and probably phase 1.5 as well all rolled into one. So if you boot from it, you will boot linux from /boot directly then do your grub-install which may just be writting the same 512 byte image into the MBR as IS ALREADY THERE. I hope this is now clear.... -- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED] https://www.redhat.com/mailman/listinfo/redhat-list
