I got a Dell PowerEdge 2850 server which was thrown out at my work(?) and I decided to use it(after all it was a free hardware). I installed initially 12.3 RC1, but because I was not able to setup the sound(and my son pressed to have the server for him) I decided to install 12.2 which I’m more familiar with and I know where to find any package that I need. After installation via a Network CD I get into emergency mode. Trying to find a solution today, i found a similar problem for 12.3 which I was thinking that will solve my problem also, but in fact my problem is different. I will provide as much info as I read from the other thread to give hopefully a clear picture of what is going on.
out of what I can see blkid list the partions to be on sda5,sda6 and sda7 but fstab has them as sdb5,sdb6 and sdb7. I don’t know why. I choose to install over the previous partitions which was the default solution at installation. Or is it? Now I’m not that sure but I just agreed with the solution provided by the installation process.
Any solutions? any other details needed? Thanks
I can’t answer your sda vs sdb question, other than to note insertion of a USB device could cause the sda to change to sdb.
wrt the emergency mode, is there any chance the radeon 7000 (?) hardware in your Dell and its compatibility with the recent ‘radeon’ driver is related to this problem ? Can you boot with a different graphic driver (such as vesa or fbdev ) ?
I could try but I don’t remember how to do add the correct parameter is it just vesa? wonder how did it work in my first try of 12.3, but that’s not important now. I will search for the parameter
You can try the boot code ‘nomodeset’ with, or without the boot code ‘x11failsafe’ (both may or may not be needed). That may bring you to either the fbdev or vesa driver (I don’t know which). Then you could examine the /var/log/Xorg.0.log file to see which graphic driver was used in the end.
Alternatively to load the vesa driver, edit the /etc/X11/xorg.conf.d/50-device.conf file (root permissions requried) to look something like:
and when rebooting also use boot with the boot code ‘nomodeset’ . The idea behind that is by specificying ‘nomodeset’ one will disable the kernel mode setting (where the kernel attempts to automatically identify the graphic mode/driver needed) and then the edits in the 50-device.conf will result in the ‘vesa’ driver being loaded.
Never use /dev/sdX directly. They are not guaranteed to be stable across reboot. Use mount by UUID, LABEL or stable links under /dev/disk.
How can I change them at installation time? Is there a way? I recall in the partition part something vaguely that the mounting by can be changed.
Should I just retry another installation? or I can change it at this point in time. Partially the system is up and I have access to yast2, even though some errors are flashing when choosing some options.
oldcpu:
I will try your suggestions first just to see if it makes any difference as soon as I get back home and post the results.
Thank you for suggestions.
I was sure mount by id is default today, so I’m a bit surprised to see your installation using /dev/sdX.
or I can change it at this point in time. Partially the system is up
Just edit /etc/fstab and replace /dev/sdb with /dev/sda. Then reboot. You should boot now. Then go into YaST2 partitning tool, edit each file system and in fstab option select how you want to mount it. UUID or device ID is good choice. Make backup of /etc/fstab and verify that new /etc/fstab looks OK before reboot.
Hi arvidjaar, well I didn’t change that to be by-name. not sure why is not by-id which would have probably avoided this isue for me. Anyway did the changes in fstab and added a nomodeset and got it to boot. After it completed the instalation it complained that it cannot display for me, but i was in a graphic mode already. Maybe not the best one but good enough to do upgrades on the packages. I will try to follow oldcpu’s notes and see about graphics and get back here if not successful, but I think the initial problem was resolved. subsequent ones should probably be under a new thread. So changing the fstab which got somehow mixed up to corrent partitions resolved the login problem. Thanks to you and to oldcpu for help/guidance.
Likely you are now using the ‘vesa’ graphic driver, which is not optimal, but at least it brings up a GUI. You can check which graphic driver is in use by looking inside the file /var/log/Xorg.0.log and it should have multple entries such as () VESA (0) or (II) VESA(0) indicating VESA graphic driver … or it could be () FBDEV (0) or (II) FBDEV(0) indicating FBDEV graphic driver. If you need help there you could copy the content of the /var/log/Xorg.0.log file and paste it on SUSE Paste , press contribute, and then post here the URL and we can tell what driver is in use.
It may or may not be possible to custom tune the 50-device.conf file to get the radeon driver to work (by adding special ‘Options’ ) but that could be a painful trial and error process.
As I said the initial problem was resolved, but subsequent ones appeared because I assume I didn’t do something because I was anxious to see if it works and because I didn’t know how to do it and didn’t try to get more info. So after I changed the fstab contents to have the drives by name changed from sda* to sdb* I tried to reboot and the system restarted and says GRUB nothing else. I tried few things and I hope that I will recall them in the correct order. First of all I created a LiveCD disk and tried to boot in and correct the things according to knurpht how to. I tried to see what changed in my system. the commands retrieved less info as it seems some things were lost.
I then tried to modify boot loader(which gives an error (Because of the partitioning, the boot loader cannot be installed properly), which let’s you move on after ok button is clicked. None of the choices for Boot Loader location are selected, so I tried to emulate another machine I have by selecting the Boot from Root partition and filled all the fields in the Boot Loader Options which were all empty emulating another system and using the UUID of the swap partition in the kernel line parameter. I modified the /etc/X11/xorg.conf.d/50-device.conf according to oldcpu suggestion and adding nomodeset to Grub parameters. I rebooted the system but it seems that because Boot loader gave an error nothing got saved and I was back to square 1. GRUB and then stop. I would appreciate any suggestions as I’m stuck at this point and not sure what to do. I can log in via LiveCD (with the errors because of Radeon card and cannot change anything via Yast. I would change the things manually in the configuration files if needed but not sure which ones, i might guess few of them needed by GRUB. Meanwhile I will read more about it. Thanks. If you think it should be a different thread then I will move it.
When running from a CD you have to mount that partition you want to work on . I think you were just looking at and changing the CD image version which is not persistent
Thank you gogalthorp. I tried again, this time following the steps described by Knurpht on how to boot from a LiveCD disk:
1. su -
2. mount /dev/sdb6 /mnt (the partition for / )
3. mount --bind /dev /mnt/dev
4. chroot /mnt
5. mount /proc
6. mount /sys
Then I open via the graphical interface Yast and in Boot Loader I get the same emptiness. I will try to fill again the info I tried earlier and see what happens until I get a reply from anyone. Having a physical LVM I tried to follow the instructions of someone who had a similar issue, but I think that he had a logical LVM and I cannot figure out some of the instructions there. So if anyone has further ideas I do appreciate the time spent on helping me.
Probably you would know that it didn’t work for me(modifying boot loader via yast). after rebooting w/o LiveCD these messages appear;
started Network Manager
started Network Manager Wait Online
started LSB: handles udev coldplug of bluetooth dongles
started Postfix Mail Transport Agent
with some coloured bars and then it hangs. I will have to boot again via LiveCD, but to do what? Any ideas? Should i attempt another installation?
whatever you exactly did, the partitions that still were listed in you first posting seem to be gone in /etc/fstab now.
So it is quite clear that you can’t access them any more.
Then, in your first posting (code there) it looks a bit like you used LVM.
Further, you write that you can’t use YaST from the live CD.
So on the basis of what remains after your changes the possibilities seem to be few now.
You could perhaps try to restore the original contents of your /etc/fstab (you have a backup of the contents of that file
in your own 1st posting in this thread), and see if that would work.
Chances are ???
Or you could start all over again, i.e. re-install, perhaps trying the installer DVD this time.
thank you for your input Ratzi. I would like to understand what happened after my first reboot. The /etc/fstab had the addresses by-name which probably got lost after the reboot. I have a physical LVM on this server and I can try to repopulate fstab to the original one and then get another try to the boot loader. I could probably try to have them by-id or uuid instead of by-name. If it doens’t work I will retry a DVD installation. Thanks again.
It is hurting that you can’t use YaST’s partitioner to change the way the hard disk(s) or partitions are accessed
(by name / by device ID etc…).
This would reduce the possibilities for typing errors and the like.
Perhaps you could as well try openSUSE 12.1, or wait for openSUSE 12.3.
My saga with openSUSE 12.2 on this server with LVM continues. So after I was unsuccessfull with the LiveCD experiment I took ratzi’s advice and tried a DVD installation.
I started the installation this time formatting the /home directory from the previous installation beside the root(/) partition and had to leave the installation unnatended. When returning I was at the emergency login message. In the fstab and fdisk I found discrepancies sda vs sdb and modified fstab to the UUID of the partitions as per /dev/disk/by-id. I added to the boot via YAST nomodeset(as per oldcpu’s advice) and rebooted. This worked fine and after reboot I had just the complain that couldn’t find /home/user/.ICEManager cannot recall exactly the file missing. I switched to a terminal(ctrl-alt-F2) and found that the home directory for the user doesn’t exist. In Yast was listed correctly. so i added another user test via YAST and was able to boot successfully. I applied some patches and when rebooted today got back to GRUB. In Partitioner I changed to have by UUID the fstab. What should I do? try to reproduce the flow and open a bug? or just fix it manually. Well I don’t even know what right now, probably GRUB as a start. Any suggestions? Thank you.
I’m still struggling to get this going. I noticed something with the partitions which might be the cause of the grief and I would like to ask someone who knows about it for an advice:
fdisk -l
Disk /dev/sda: 1199.9 GB, 1199906488320 bytes
255 heads, 63 sectors/track, 145880 cylinders, total 2343567360 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x00000080
Device Boot Start End Blocks Id System
/dev/sda1 63 192779 96358+ de Dell Utility
/dev/sda2 192780 8594774 4200997+ 8e Linux LVM
/dev/sda3 * 8595456 2343567359 1167485952 f W95 Ext'd (LBA)
/dev/sda5 8597504 12804095 2103296 82 Linux swap / Solaris
/dev/sda6 12806144 54749183 20971520 83 Linux
/dev/sda7 54751232 2343546879 1144397824 83 Linux
I don’t know why this W95 partition exists. Initially when I found the server it had a RedHat 4 on it. I installed OpenSuse 12.3 and I selected to format whatever was on the disks. I didn’t make any other changes and not sure what and why that extended partition of W95 is there. My question is do I need it? also do I need the Dell Utility partition? It might be that some tools are on that partition but I have no clue and I wouldn’t know how to access it. I’m convince that sda3 is messing my installation because I selected to have the system boot from sda6 and now I can see that sda3 was selected as boot(*). So to reiterate, I don’t have anything that I need as far as I know on any partitions. Should I select to have them removed before installation? Thank you for any advice/suggestion.
Something in your code pasting is skewed, had to edit it.
> Code:
> --------------------
>
> fdisk -l
> Disk /dev/sda: 1199.9 GB, 1199906488320 bytes
> 255 heads, 63 sectors/track, 145880 cylinders, total 2343567360 sectors
> Units = sectors of 1 * 512 = 512 bytes
> Sector size (logical/physical): 512 bytes / 512 bytes
> I/O size (minimum/optimal): 512 bytes / 512 bytes
> Disk identifier: 0x00000080
>
> Device Boot Start End Blocks Id System
> /dev/sda1 63 192779 96358+ de Dell Utility
> /dev/sda2 192780 8594774 4200997+ 8e Linux LVM
> /dev/sda3 * 8595456 2343567359 1167485952 f W95 Ext’d (LBA)
> /dev/sda5 8597504 12804095 2103296 82 Linux swap / Solaris
> /dev/sda6 12806144 54749183 20971520 83 Linux
> /dev/sda7 54751232 2343546879 1144397824 83 Linux
>
>
>
> --------------------
>
>
> I don’t know why this W95 partition exists.
It is not a Windows partition. It is an extended or container partition
in classical style, also known as W95. Many people have it. If you erase
it, you also erase sda5, sda6 and sda7, that appears to contain Linux.
> wouldn’t know how to access it. I’m convince that sda3 is messing my
> installation because I selected to have the system boot from sda6 and
> now I can see that sda3 was selected as boot(*).
You can not boot directly from sda6, sda3 will be picked instead. You
can not mark a logical partition as bootable.
–
Cheers/Saludos
Carlos E. R. (12.3 Dartmouth test at Minas-Anor)