A client recently changed their UPS and now claims that their server is not coming back up after am extended power outage. They have tested their system and it will keep the last state, so if the power is removed while on, it will automatically power up again.
The question is, in Linux you can request a halt so that when power returns it will restart, but how do you do the same in Windows Server?
I have a Gigabyte GA-965P-DQ6 motherboard with the latest BIOS flashed in (version F13).
I also have 4x sata disks (in proven working condition) from a different server. Two of them are Western Digital ATA WDC WD1003FBYZ-0 and the other 2 the (very similar) ATA WDC WD1003FBYX-0.
To connect the disks I used 2x ASMedia ASM1061 PCIe-1x to SATA sata controllers (firm.ver. 3.80), since I have already used up all of my motherboard's SATA and PATA ports on other hard disks.
The problem is that once I connect the disks onto the sata controllers, the boot process stops sometime before lilo (I am running slackware linux) boot loader menu is loaded. In fact the boot process gets stuck in the following screen:
If I remove the disks, but keep the controllers connected onto the motherboard, the boot will continue normally. Then once booted I can "hot plug" the disks and they will show up fine in the system (using dmesg to verify). I have used cfdisk to overwrite the "MBR" of each disk and to make sure that the bootable flag is not set.
I have also double checked whether the BIOS uses as the first boot device the disk I actually want (which is not part of these 4 disks) and indeed I can verify that this is the case. Here you can see that the BIOS has identified the following boot devices (disks and pci cards):
You can see here most of the disks that are directly attached to the motherboard's SATA ports - so none of the 4 disks that create the problem. You can also see that the last option #8 "Bootable add in cards" refers to the sata controller cards. So on top there is the preferred boot device and down at the bottom there is the sata controller cards, but somehow during boot the process will stop.
So I am out of ideas about what could be causing the boot process to halt. I have not been able to find info about how to log on the sata controller's boot configuration menu (if any exists): no informative message about a key combination that would allow that appears, nor I have found any relevant info on the internet. Is it possible that this sata controller model does not have a boot configuration screen?
EDIT: Thinking out loud, I wonder if the normal procedure when you try to boot a non bootable disk would throw a "boot failure" message. Since this does not happen here maybe this is not at all a problem with the extra disks trying to be booted from, but something different altogether (no clue as to what though).
EDIT:
relevant contents of fstab per comment request (I wonder about the usefulness though since this is a BIOS problem):
/dev/md1 swap swap defaults 0 0
/dev/md0 / reiserfs defaults 1 1
halt can turn off the machine,
but shutdown now doesn't turn off ,it just logs the root off.
Anyone knows the reason?
I have had a hard drive failure which has caused a lot of processes to go into an uninterruptable state ("D" state in ps- See this question). I cannot kill these processes, so I wish to shut down, but unfortunately halt is also now showing as "D" and the machine will not restart!
ben@jack:~$ ps aux | grep "halt"
root 7875 0.0 0.0 4148 700 ? S Mar08 0:00 /bin/sh /etc/rc0.d/S01halt stop
root 7880 0.0 0.0 10436 680 ? D Mar08 0:00 halt -d -f -i -p
How can I recover from this?
EC2 Ubuntu servers erase all disk contents when being shut down. Following an unfortunate accident, I have decided to prevent the command-live halt, poweroff and shutdown.
What's the best way to do it? I thought about renaming these commands (at /sbin) to something like HALT_RENAMED___ERASES_ALL_DISK_CONTENTS.
Are there any files, other than the three listed above, that needs to be handled?
I've noticed that halt and poweroff are merely links to reboot. Should reboot be renamed, too?
Adam