Upgrade from 13.1 to 13.2 64 bit gone wrong, will not boot.

English Install/Boot/Login
1

Hi I’ve been trying to get to the bottom of this problem. I was trying to do an update upgrade of 13.1 to Leap 42 and started following the instructions…

First I made sure my 13.1 was up-to-date.
Then I switched to the basic upgrade repos and turned off the rest.
It stumbled a couple of times during the download/install of 13.2 packages.

So, having got that far, I thought I had everything good to reboot and took the plunge…and wouldn’t reboot.

Here’s a bit of background.

I had 2 x 2TB physical drives partitioned sda1/sda2/sda3/sda4 and sdb1/sdb2/sdb3/sdb4 running a softward RAID1.

sda1 & sdb1 - md126 - ext4 101MB boot partition
sda2 & sdb2 - md124 - swap 2.0GiB swap
sda3 & sdb3 - md125 - ext4 20GiB root
sda4 & sdb4 - md127 - ext4 1.8TiB /home

Now the first problem that was causing the set-up not to boot was that the updating of the kernel in the boot partition didn’t quite go right. If my reading is correct, the 100MB boot partition is a bit small these days.

So I ended up creating a new problem, by trying to solve the first one. I’ve now got the drives setup like this:

sda1 & sdb1 - md126 - ext4 512MB boot partition
sda2 & sdb2 - md124 - swap 1.6GiB swap
sda3 & sdb3 - md125 - ext4 20GiB root
sda4 & sdb4 - md127 - ext4 1.8TiB /home

These appear to be functioning as a RAID1 setup (both partitions in every array showing up UU)

fdisk -l:
https://farm6.staticflickr.com/5831/22823902020_153757a969_z.jpg](https://flic.kr/p/ALSsU5)https://farm1.staticflickr.com/782/22725595937_c2111503cf_z.jpg](https://flic.kr/p/ACbBVB)
IMG_20151119_144204_873 by macplaxton, on FlickrIMG_20151119_144211_788 by macplaxton, on Flickr

cat /proc/mdstat:
https://farm6.staticflickr.com/5643/22725688758_4647869a40_z.jpg](https://flic.kr/p/ACc6vY)
IMG_20151119_145832_336 by macplaxton, on Flickr

/etc/fstab:
https://farm6.staticflickr.com/5655/23155645401_4c09183393_z.jpg](https://flic.kr/p/BhbJDx)
IMG_20151119_144617_261 by macplaxton, on Flickr

blkid (See next post)

I can’t seem to install bootloader grub2 from the main DVD iso image (on bootable USB). I can’t seem to do it either from the rescue prompt using yast. I’ve edited my fstab manually, but I may need to do it again.

Anyone got a fresh pair of eyes to help? I’ve kind of lost my place with this.

2

blkid:
https://farm1.staticflickr.com/725/22751798009_54ca8b0655_z.jpg](https://flic.kr/p/AEuUTt)
IMG_20151119_144300_715 by macplaxton, on Flickr

3

And this is where the problem started:

https://farm1.staticflickr.com/625/22823898150_7c2ca34b8c_z.jpg](https://flic.kr/p/ALSrKm)
IMG_20151113_132038_682 by macplaxton, on Flickr
https://farm1.staticflickr.com/574/23155649471_a44d3da489_z.jpg](https://flic.kr/p/BhbKRH)
IMG_20151115_210358_938 by macplaxton, on Flickr

(the UUID may have changed since I shot that, but it was pointing to the /boot partition on the drive after growing the size of it)

4

Are you using legacy BIOS or EFI? How exactly do you install grub (please paste exact command and its output)?

Regarding moutn error - something misinterprets UUID with colons for NFS reference. I briefly tested in Leap - plain “mount” does not have issues with it, it correctly mounts MD. So it sounds like YaST2 issue. If you can describe what have you done step by step so developers can reproduce it, it makes sense to open bug report. In the mean time you can use UUID or LABEL in /etc/fstab to work around this.

5

Thank you for your reply.

Legacy BIOS I think, as I’ve had this set up working okay with 13.1 and earlier.

I don’t think it’s a bug to start with. The initial point it went wrong was the 100MB separate boot partition was too small by a hair. So I shrank the swap and expanded the boot. I’m worrying about the partition table now. In rescue mode, I’ve done:

https://farm6.staticflickr.com/5773/22728737308_3397bc7381_z.jpg](https://flic.kr/p/ACsHKb)
IMG_20151119_171410_455 by macplaxton, on Flickr

https://farm6.staticflickr.com/5753/22755450349_c0d803bb00_z.jpg](https://flic.kr/p/AEPCAR)
IMG_20151119_172922_582 by macplaxton, on Flickr

https://farm6.staticflickr.com/5794/23133752262_5dfe4949bb_z.jpg](https://flic.kr/p/BffwzC)
IMG_20151119_174945_849 by macplaxton, on Flickr

6

No, it is impossible to install grub2 on Linux MD device. You need to install it on MBR of physical drive(s). It is possible to install it several times in MBR of different physical drives; actually YaST supports it and offers when it detects /boot on RAID1.

7

Ok, I’ve got this far:

I’ve installed grub2 into the MBR of each drive, so regardless of which one starts first, or if one drive fails, it should start…

https://farm1.staticflickr.com/690/23159834581_68584eb997_z.jpg](https://flic.kr/p/BhycWM)
IMG_20151119_181020_893 by macplaxton, on Flickr

Now I’m dumped at this bit:

https://farm6.staticflickr.com/5780/22525547914_8d9a0696e2_z.jpg](https://flic.kr/p/AjvjBL)
IMG_20151119_181505_168 by macplaxton, on Flickr

8

Because openSUSE can have at any given time 3 complete kernels and associated boot files you need at least 4oo meg and probably 500 meg to be safe 100 meg is far to small to handle multiple kernels

9

grub-install only installs GRUB binaries. It does not create or modify grub.cfg; without grub.cfg GRUB loads and goes to command line.

Also by default GRUB assumes grub.cfg is located in /boot/grub2. If you were doing this from within rescue boot, /boot/grub2 is not referring to your “real” /boot as so installed binary looks in the wrong place.

I’m afraid I do not quite understand what your configuration is now, but assuming you have openSUSE installed you can do usual


mount */your/root/device* /mnt
mount */your/boot/device */mnt/boot
mount --bind /sys /mnt/sys
mount --bind /dev /mnt/dev
mount --bind /proc /mnt/proc
chroot /mnt
grub2-install /dev/sda
grub2-mkconfig -o /boot/grub2/grub.cfg

Adjust as suitable for your configuration (e.g. I do not know if you have separate /boot or not).

10

Thanks for the explanation, I’ve been kicking the can down the road on that one and only just got round to making the boot partition bigger only to completely break things.

11

Understood, I think there may be a config file, but it’s wrong or not working.

I was aware of that, and had entered the code as above and mounted all the root first ( /dev/md127 etc. ), followed by the separate boot partition, binding things up and chrooting.

I’ve since used commands in the grub shell and got a partial boot, but it seems to be failing to mount things properly. I’ll get details up later.

12

From the Grub2 shell, I set the root for grub, loaded the kernel and initrd and booted

https://farm6.staticflickr.com/5758/23114671366_e0e8cc3b3f_z.jpg](https://flic.kr/p/BdyJuE)
IMG_20151119_222331_064 by macplaxton, on Flickr

It starts off okay, but it stalls around here:

https://farm6.staticflickr.com/5802/23151162312_096b496810_z.jpg](https://flic.kr/p/BgMKYU)
IMG_20151119_222442_368 by macplaxton, on Flickr

Then it dumps me at an emergency prompt:

https://farm1.staticflickr.com/570/22746627907_0fc39e1983_z.jpg](https://flic.kr/p/AE3pZM)
IMG_20151119_222531_055 by macplaxton, on Flickr

Going back a page to see the error:

https://farm1.staticflickr.com/672/22844960480_16f3aaee6c_z.jpg](https://flic.kr/p/ANJoQY)IMG_20151119_222545_185 by macplaxton, on Flickr

13

It stops because device /dev/disk/by-id/md-uuid-xxx:255b63c8 is missing, but that makes no sense - this UUID belongs to filesystem not to Linux MD device. At least, as long as we can believe blkid output shown earlier [1]. Did you edit /etc/fstab manually? Then you need to fix it. Either use Linux MD UUID and /dev/disk/by-id or use filesystem UUID and /dev/disk/by-uuid (or simply UUID=xxx).

[1] It would make comparing, quoting and helping you much easier if you simply pasted text output instead of huge images. It means we can only compare visually and it is easy to make mistakes. And I have no way to quote them and demonstrate what I mean, nor copy-paste them in my reply.

14

Thanks arvidjaar, I did edit the /etc/fstab file manually, as the UUIDs of the boot and swap partitions had changed due to amendments. I will go back and check through them carefully and see.

[My apologies for the large pics. I can understand yours and others view on the shortcomings of it. It was a clumsy workaround to the problem of conveying lots of screen info accurately. I’ve not the faintest idea, bar typing and proof-reading everything on screen how I can simply copy and paste text output between a non-booting openSUSE box on the one hand, and this Windows 7 laptop on the other. I can managed copy/paste from a terminal window to a forum post on a booted machine. For future reference (if it is quite simple to do) could you explain how to do so?]

15

Further to this, I’ve edited the fstab to the correct UUIDs as something had changed since the last manual edit. Before editing I now had five lines instead of four, with two lines referring to swap areas.

So having done that and updated the grub.cfg file with the grub2-mkconfig -o /boot/grub2/grub.cfg, I’ve taken a step backward as the grub2 shell only lists files in the /boot partition. It doesn’t appear to have mounted anything else.

I notice prior to it timing out previously, it does mount a swap and root, but by disk sdx UUID rather than mdx UUID. Is that relevant?

16

It i better to use UUID rather then the devices address. the SDX# nomenclature can change ie it is not entirely stable it comes from the BIOS and it may change depending on what is plugged into where. UUID is stable in that a pseudo random number is used when the partition is created with a great probability of being unique. In the /dev/disk directory you will see the files representing the UUID and other ID methods. They are simply links back to the sdX# devices that has that ID

17

Just to clarify, I’m now back to partial boot.

It’s still stalling when it comes to mount the drives. What I meant in my previous post, was that it appeared that a UUID of a physical drive partition was being mounted, instead of that of the corresponding RAID partition.

When I say stalling, it says

(1 of 3) A start job is running for dev-disk-by\x2did-md\x2duuid\x2d[UUIDs from fstab].device (1min/ 1min 30s)

I’m going to go back to check the partition table as I’m not happy with these “unknown partition table” mentions in the log with regards to a couple of the RAID 1 partitions.

18

I do not understand what you mean. Please paste output of blkid and mark UUIDs you mean.

19

blkid output (typed out from screen):

tty1:Rescue:˜ # blkid
/dev/sda1: UUID="8ac23086-a200-bdba-c4b4-2555bf0c8fe2" UUID_SUB="3adb04ea-3331-3ee9-f07e-4a5e655b9c67" LABEL="debian:125" TYPE="linux_raid_member" PARTUUID="00051cf7-01"
/dev/sda2: UUID="a58fa429-8414-d79c-2179-2cae28522ece" UUID_SUB="8deaf4df-088f-a5f1-8dc4-314e3aae2c04" LABEL="debian:124" TYPE="linux_raid_member" PARTUUID="00051cf7-02"
/dev/sda3: UUID="d315edee-e46b-c575-bb12-bd2bbb9d07f5" UUID_SUB="82598464-0c97-4845-a87f-77062da1904f" LABEL="linux:2" TYPE="linux_raid_member" PARTUUID="00051cf7-03"
/dev/sda4: UUID="e355c9a0-8f6f-2255-3d17-60b2ccc304de" UUID_SUB="05a3ee32-b067-a524-d579-ce0caaefef6b" LABEL="linux:3" TYPE="linux_raid_member" PARTUUID="00051cf7-04"
/dev/sdb1: UUID="8ac23086-a200-bdba-c4b4-2555bf0c8fe2" UUID_SUB="2b8390be-6d10-d39a-2e07-5f87480a08cb" LABEL="debian:125" TYPE="linux_raid_member" PARTUUID="000cac99-01"
/dev/sdb2: UUID="a58fa429-8414-d79c-2179-2cae28522ece" UUID_SUB="8052d31d-8741-0ca3-b455-2eeb4db6b6fc" LABEL="debian:124" TYPE="linux_raid_member" PARTUUID="000cac99-02"
/dev/sdb3: UUID="d315edee-e46b-c575-bb12-bd2bbb9d07f5" UUID_SUB="fa2b44e5-bacd-5913-70c3-9de5b3be4899" LABEL="linux:2" TYPE="linux_raid_member" PARTUUID="000cac99-03"
/dev/sdb4: UUID="e355c9a0-8f6f-2255-3d17-60b2ccc304de" UUID_SUB="29cce7ca-1a8e-2296-e0d6-0a163852e630" LABEL="linux:3" TYPE="linux_raid_member" PARTUUID="000cac99-04"
/dev/loop0: TYPE="squashfs"
/dev/loop1: TYPE="squashfs"
/dev/loop2: TYPE="squashfs"
/dev/loop3: TYPE="squashfs"
/dev/loop4: TYPE="squashfs"
/dev/md127: LABEL="boot" UUID="72215086-e227-485a-8460-75be5f99c4bf" TYPE "ext4"
/dev/md126: LABEL="swap" UUID="0fc5510a-a178-4389-9f4f-986c0122210e" TYPE "swap"
/dev/md125: UUID="afa31a78-fb19-4e92-be00-2177f16d09ec" TYPE="ext4"
/dev/md124: UUID="bbe5e47d-21a8-4c2c-a59d-68ff255b63c8" TYPE="ext4"
/dev/sdd1: SEC_TYPE="msdos" UUID="5F67-0F78" TYPE="vfat" PARTUUID="10ce3e4f-01"
/dev/sdd2: UUID="2014-10-27-14-55-49-00" LABEL="openSUSE-13.2-DVD-x86_640051" TYPE="iso9660" PARTUUID="10ce3e4f-02"

it was finds a device:

 OK ] Found device /dev/disk/by-id/md-uuid-8ac23086:a200bdba:c4b42555:bf0c8fe2.
       Starting File System Check on/dev/disk/by-id/md-uuid-8ac23086:a200bdba:c4b42555:bf0c8fe2...

which is sda1/sdb1 (swap) rather than uuid-0fc5510a:a1784389:9f4f986c:0122210e which is my RAID1 swap.

Here’s my fstab:

Rescue:/ # cat /etc/fstab
/dev/disk/by-id/md-uuid-0fc5510a:a1784389:9f4f986c:0122210e swap     swap   defaults          0   0
/dev/disk/by-id/md-uuid-afa31a78:fb194e92:be002177:f16d09ec /        ext4   acl,user_xattr    1   1
/dev/disk/by-id/md-uuid-72215086:e227485a:846075be:5f99c4bf /boot    ext4   acl,user_xattr    1   2
/dev/disk/by-id/md-uuid-bbe5e47d:21a84c2c:a59d68ff:255b63c8 /home    ext4   defaults          1   2
20

You misunderstand blkid output or use confusing terminology. Your MD UUID is 8ac23086-a200-bdba-c4b4-2555bf0c8fe2 while 0fc5510a-a178-4389-9f4f-986c0122210e is UUID of content of MD device. It can be swap, or it can be filesystem. Or may be something else like LVM PV. But it is not UUID of MD device.

Here’s my fstab:

Rescue:/ # cat /etc/fstab
/dev/disk/by-id/md-uuid-0fc5510a:a1784389:9f4f986c:0122210e swap     swap   defaults          0   0

As I already told you it is wrong. /dev/disk/by-id/md-uuid-XXX is created using Linux MD UUID while your fstab entry is using swap UUID. Device that you put into /etc/fstab does not exist and cannot exist (as long as we believe that UU in UUID really means Universally Unique).