Friday, July 20, 2012

Recovering from SecurePlatform Boot Failure


DRP Experience from Joe for SPLAT Boot Failure: 
We recently assisted a client that had a failure with one of two Check Point SecurePlatform Security Gateway cluster nodes where after applying a license the machine failed to boot. The issue was that the machine sat at a “Loading” prompt after attempting to boot from CDROM (unsuccessfully), and then booted from C:.  It would appear the boot loader ‘grub’ was not starting correctly, and likely the MBR was corrupted.  The machine never got past the “Loading” message.
As we have seen this once before recently we thought it would be worthwhile posting a quick ‘how to’ to recover from this.  Note: This issue was encountered on an HP ProLiant DL380G6 platform.
To fix you need to boot into a rescue mode of sorts and re-install the boot loader/MBR:
  • Boot from the SPLAT R70 ISO and allow the install to continue to the first installer “welcome” screen [at the ‘press any key after 90 seconds’ – press a key].
  • Press Alt+F2 to change to the “install shell” – basically a bash shell that runs with the installer.
  • At the command prompt, make a temporary directory to mount the real disk partitions:
    • mkdir /splat
    • mkdir /splat/boot
    • mkdir /splat/tmp
  • Mount the real disks:
    • mount /dev/cciss/c0d0p7 /splat
    • mount /dev/cciss/c0d0p1 /splat/boot
    • mount /dev/cciss/c0d0p5 /splat/tmp
  • Copy the ‘dev’ entries into the /splat/dev directory [needed to reinstall the MBR]
    • cp -r /dev/cciss /splat/dev/
  • Run up a chroot
    • chroot /splat /bin/bash
  • Run the ‘grub’ utililty
    • grub
  • This will run up a management interface for use in editing the boot loader config – grub is the bootloader.
  • Configure grub to use the correct root disk, enter:
    • root (hd0,0)
    • setup (hd0)
  • Reboot and that should do it.
Note: For reference, “hd0” maps to a physical disk in the /boot/grub/device.map file, e.g. below, so should be possible to transpose over to another box, such as Dell or IBM [which don’t use the cciss disk driver]
[[email protected]]# cat /boot/grub/device.map
# this device map was generated by anaconda
(fd0)     /dev/fd0
(hd0)     /dev/cciss/c0d0
"

No comments: