I wonât say I am doing any good job of diagnosis, because to do that I would had to stepback every time after I got past a snag, and verify or prove my theory of discovery, but instead of that, I just simply past whatever snag I got past, and roughly guess the best logical reason of it and share without really recreate the issue or prove the solution.
My main snags were these:
GRUB not original SuSE, because I am multi-booting Kubuntu Fedora Core & some other experimental stuffs. My GRUB is GRUB 2.x while SuSE is GRUB 1.x - this is a relatively small issue, but can bug new users badly.
Soft RAID not starting not mounting during boot. This is a big issue to me, my /home & /home/user/raid are are soft RAID0, it is still in some trouble although I can already boot straight into runlevel 5, the pain now is it will still not straight away see my raid volume, and it wait there for 20sec unlike in other Linux, it displays my RAID uuid and says awaiting for it to appear, and show a long bunch of dots⌠after 20s it would see my raid and continue. I forced** /etc/init.d/boot.md** to early trigger in /etc/init.d/boot.d. I added these to GRUB2** /boot/grub/grub.cfg** which is the file that took the role of** /boot/grub/menu.lst**
xorg driver issue, somehow ATI radeon driver supplied inside SuSE was unable to get me to GUI level. I must install driver from ATI site, ati-driver-installer-11-3-x86.x86_64.run this version works for me. I think my own xorg.conf is abit too complicated for SuSEâs defaults, I have dual screen and rotated., with Kubuntu & FC I had only a little trouble with xorg.
menuentry "SuSE 11.4, with 2.6.37.1-1.2-desktop (on /dev/sdf6)" {
set gfxmode=1680x1050
set gfxpayload=keep
insmod raid
insmod mdraid
insmod gfxterm
insmod vbe
insmod part_msdos
insmod ext2
set root='(/dev/sdd,msdos5)'
search --no-floppy --fs-uuid --set 61a9e3b8-09c0-42bd-adf1-ce4b152f8411
linux /vmlinuz-2.6.37.1-1.2-desktop root=UUID=a9a93efe-4e1f-455a-bf65-0ecfe93838ad splash=silent showopts
initrd /initrd-2.6.37.1-1.2-desktop
}
If you checked the above codes there are raid related modules and display splash screen addressing statements in GRUB2 added by me. I just added them and got past my snags, however what I did not do is to try remove them one by one, and see if I had actually added unnecessary (over kill) settings or not. That is to say I am still not optimized, I am just out of the woods alive.
I also have xdm snags, when it was in unpredictable state in parallel boot mode, I used combinations of followings to get my GUI, but surely, none of it will help before I installed the above-mentioned ATI Linux driver. The pain in the Azz of this driver is each & every time SuSe kernel is updated, you have to reinstall it again, and before you do so you got no GUI once you reboot. So users are hereby warned!
sudo init 5 (when you are thrown to runlevel 3 - runlevel command will tell you)
sudo kdm restart (I use KDE)
sudo service xdm restart
Any of the above, may work or a combination of the above, had been able to get me into GUI, that was when parallel boot was making things unstable for me, and I was shooting in the dark. After I rid parallel boot, it became stable, and init 5 was good enough, only in one occasion, I needed sudo service xdm restart but I think that instant does not count because I was then tweaking runlevel, and it might had been my own tweaks that got myself into troubles.
LAN NIC lost, no networks, if your /etc/fstab for example have cifs samba or nfs mounts, your boot issue will be even harder due to this, remark away these stuffs in /etc/fstab until after you are out of the dark with the boot issues. Otherwise, you can not isolate issues.
The other snag is vmware.com virtual machine services did not start, and alsa sound server did not start, after 1st removing parallel boot and tweaking runlevel services they worked, in the process of experiment I broke sound in one installation, still donât know how to fix it, but I am using the other installation now. :X If you had more harddisk ready for partitioning, and you know how to properly setup multi-boot, you can trouble-shoot and tweak in safer conditions, by playing with more than 1 instant of SuSe installation. I also used virtual machine to mount my installation to help myself get access to the unbootable deadlock situations, live CD can surely help you as well.
There are many more minor snags like my konsole hostname got lost, and some un-useful daemons reporting errors during boot, postfix, mysql, brill codes etc, can all come in and complicate your troubles, knock these services & daemons away, and focus on getting GUI display and LAN NIC up, and your /home partitions can mount so that you can use a comfortable GUI environment to deal with the rest of the issues. Otherwise, you are unable to isolate issues, you need to single out these issues one by one.:\
My old RedHat 8 days experience helped, because in those days, we needed to learn how the links inside /etc/init.d/boot.d and /etc/init.d/rc5.d are meant to do.
they are alike ON/OFF switches to each of the services of linux. They are like timer sequencers, K01xdm will kill xdm very first K99ssh will kill ssh last and so on the 2 digit numbers are sequence order, and letter K means kill S means start. So in each runlevel, these links there programs all the services that you have, weather they will kill which 1st and which next, and after killing all, which to start 1st, because of interdependency with each other you can not successfully start / stop in illogical sequence, even if you needed the service, you are better of to kill it off 1st and start it a while after, to have a clean restart or reset effect that will ensure things work. All your services are supposed to be listed in /etc/init.d and in each of runlevel e.g. /etc/init.d/rc5.d for level 5 which is your GUI login level, you only link to these services listed in /etc/init.d.
For example if you mount nfs or samba over network, you surely have to have nfs & smb turned off 1st before turning off boot.ipconfig, if you did the reverse it hangs because smb & nfs can not sign off or unmount when you flipped off tcp/ip switch. To turn back on your sequence is opposite, boot.ipconfig have to be 1st on otherwise you donât expect to be able to turn on nfs or smb.
The main problems or crisis arise here, is that parallel boot, did not run these ON/OFF in a logical enough sequence, thus hang!>:( that is due to fast processors doing multi tasking and brunch prediction, which tends to do things ahead of schedule, e.g. you programed boot.ipconfig to be killed in sequence #60 and killing smb in sequence #50 it will look logical in your rc3.d for example, but CPU multitasking executed sequence #60 ahead of #50 during parallel boot. I have no evidence, I did not try to prove it, but my hardware knowledge as a multimedia hardware designer preempted me to expect these sort of pain in the Azz all the time, in multimedia hardware, because machines needs to be in some sync and sequence, while multitasking was made to do things fast and ahead of any slow events. So I guess my theory is right, I tweaked with more safer delays between inter-dependent services. I got out of the woods.
sweatingâŚ
I use to struggle with booting up primitive prototype hardware that we designed and debug them tediously, so I applied some of these techniques to survive.
