I just did my first run on sles11sp4 → sles12sp1 upgrade.
My host is a HP hw box, and I booted on the sles12sp1 dvd, select Upgrade. I get to the screen to “Select system to upgrade”, which does not shown anything - tick at “Show all partitions” shows a lot of partitions (the host being a hypervisor host).
My disk layout for the host is:
[CODE]Model: ATA ST2000DM001-1CH1 (scsi)
Disk /dev/sda: 2000GB
Sector size (logical/physical): 512B/4096B
Partition Table: msdos
Number Start End Size Type File system Flags
1 1049kB 4294MB 4293MB primary linux-swap(v1) raid, type=fd
2 4294MB 541GB 537GB primary ext3 boot, raid, type=fd
3 541GB 2000GB 1459GB extended lba, type=0f
…
[/CODE]
So the / is on a linux raid build from sda and sdb. But does not seem to be recognized?
So the / is on a linux raid build from sda and sdb. But does not seem to be recognized?
just for good measure, have you tried switching to a text-mode console and check the MD status?
BTW: Going from SLES11SP4 to 12SP1 directly is no supported upgrade path, you’d have to go to SLES12SP0 first. But I won’t put my bet on it that this is the root cause in your situation.
In case your MD is (for whatever reason) not set up automatically, I’d suggest to switch to console right at the first installer screen (license agreement, iirc), set up md manually and then try to continue the (upgrade) installation.
So the / is on a linux raid build from sda and sdb. But does not seem to be recognized?
just for good measure, have you tried switching to a text-mode console and check the MD status?
BTW: Going from SLES11SP4 to 12SP1 directly is no supported upgrade path, you’d have to go to SLES12SP0 first. But I won’t put my bet on it that this is the root cause in your situation.
In case your MD is (for whatever reason) not set up automatically, I’d suggest to switch to console right at the first installer screen (license agreement, iirc), set up md manually and then try to continue the (upgrade) installation.
Regards,
Jens[/QUOTE]
Hi Jens,
thanks for replying.
No, I didn’t attempted to verify from console - I guess I assumed the installer to show, what is available:)
For checking and setting up: Do you have mdadm in mind?
For the upgrade path - I understand https://www.suse.com/documentation/sles-12/book_sle_deployment/data/sec_update_sle12.html as sp3->sp1 should be fine. Am I wrong, do you think?
[QUOTE=clausbc;31646]Hi Jens,
thanks for replying.
No, I didn’t attempted to verify from console - I guess I assumed the installer to show, what is available:)
For checking and setting up: Do you have mdadm in mind?[/QUOTE]
Trust no-one
A first glimpse can be had by running “cat /proc/mdstat”, and if you don’t see expected results (i.e. a single MD-RAID, level 1, consisting of /dev/sda and /dev/sdb or whatever your setup is) you could try to get things started via mdadm manually. If that fails for whatever reason, you should received hints on what the root cause may be.
If OTOH manual RAID assembly works, then you can either continue the upgrade and hope for the best, or try to debug why the array wasn’t assembled automatically. Even though I’ve had my fair share of MD-RAID experience, I’d go for the former route (or maybe because of that experience? ): Maybe it’s some code difference between the “old” MD and the one in 12SP1 that’s causing the problem - and manually assembly the RAID set may likely have cured that.
If the RAID set is already available and active, but still doesn’t show in the installer, you might want to open a service request. Debugging the installer is nothing for the faint of heart.
I read that part of the docs the same way you do - it’s just that up to now, I’ve been told that the supported upgrade path is only without skipping intermediate service packs. OTOH, you could point at the docs in any support case
And as I wrote in my first reply - IMO this should not be the problem.
opening a service request will probably get you the best result.
Now that my neck recovered from looking at the details :D, I re-read the whole thread and have the following comments:
I get to the screen to “Select system to upgrade”, which does not shown anything - tick at “Show all partitions” shows a lot of partitions (the host being a hypervisor host).
I assume you are trying to update the host, not a guest.
The various MD devices are probably handed to the guests 1:1, which is why the host sees the installed systems on these devices/sub-partitions.
(fdisk list of partitions, upgrade screen)
fdisk shows sd{a,b}2 as bootable partitions (RAIDed again), which are used to assemble /dev/md125. In the list of upgradable partitions there’s the selected entry, which has an MD name (/dev/md/linux:7) instead of an MD device partition (dev/mdp) - probably that’s the device you mount as the host’s root fs… You may want to check either manually from the installation environment’s shell, or by looking at the setup of the booted system.
On a hypervisor host, disk layouts can be pretty confusing for installers…
From SR I got the correct selection and the upgrade basically succeeded. But a number of issues followed:
Unable to install and run XEN
Very slow server in general
Unable to remote access
just to mention the worst. So I took the advice of you of the supporters - a clean resinstall of the host with backup of the DomU xmls. Everything seems fine now.
): Maybe it’s some code difference between the “old” MD and the one in 12SP1 that’s causing the problem - and manually assembly the RAID set may likely have cured that.