On Fri, Sep 6, 2013 at 6:45 AM, Darac Marjal <mailingl...@darac.org.uk>wrote:
> On Fri, Sep 06, 2013 at 02:29:55PM +0200, Robin Kipp wrote: > > Hi list, > > I just purchased an HP ProLiant Micro Server G2020T. As for the hard > drives, I installed 4 3TB Western Digital HDs. So far so good, but volumes > with a capacity greater than 2TB require a GPT partition table. > Unfortunately, the server does not support UEFI, and thus can't boot from > GPT partitions natively. > > So, when I ran the Debian installer, I used the following partitioning > scheme on all drives since I wanted to combine them in a software RAID: > > > > 1MB BIOS Boot Partition (BBP) for GRUB > > 512MB dedicated /boot partition > > partition with all the remaining disk space. > > > > I then proceeded to setup software RAID: > > no RAID on the 1MB BIOS boot partition (not sure if this is correct) > > RAID1 for the 512MB /boot partition including all the HDs. > > RAID5 for the large partition that remained for file storage. > > I then set up the /boot partition (/dev/md0) to contain an EXT3 file > system and also configured the mount point to be / boot. > > For the large partition, I setup LVM and created logical volumes for the > root and SWAP partition. I also configured those partitions accordingly so > the installer would know how to use them. > > Once I finished, the installation went through without any problems. > After the system was installed, I used the 'Install the GRUB boot loader' > option to install GRUB on all HDS (/dev/sda through /dev/sdd), which worked > just fine. > > However, when I rebooted the system I got an error message saying the > root file system could not be mounted. I suspected the LVM to cause issues, > so I re-installed everything but this time without LVM. Unfortunately, the > same issue persists… Has anyone here ever been in a similar situation and > could suggest a fix? I have a feeling I may be missing something important, > but just can't find the right path to take… > > If you're getting an error saying root couldn't be mounted then I'm > assuming that: > - BIOS has found GRUB > - GRUB has found the kernel > - the kernel has booted BUT > - the kernel couldn't find the rootfs, so therefore couldn't start init > > If that's the case, try adding "rootdelay=30" to your kernel command > line (the best way is to append it to GRUB_CMDLINE_LINUX in > /etc/default/grub and re-run update-grub", but you can temporarily add > it at grub's command line editor). > > "rootdelay" should cause the kernel to wait a few moments for all drives > to become ready, the raid to assemble and so on, before it tries to > mount rootfs. > > I have had problems with the initrd not having the LVM modules loaded in it. I had to make sure LVM was installed then run: update-initramfs -u -k all then : update-grub to get it to play nice with the system. Other than that I would just make sure your /etc/fstab file is correct. -- Shane D. Johnson IT Administrator Rasmussen Equipment
