Bug 959173

Summary: System does not boot on HP DL380 G5 Server
Product: [openSUSE] openSUSE Distribution Reporter: Ulrich Windl <Ulrich.Windl>
Component: InstallationAssignee: E-mail List <yast2-maintainers>
Status: RESOLVED NORESPONSE QA Contact: Jiri Srain <jsrain>
Severity: Major    
Priority: P5 - None CC: ancor, jreidinger, Ulrich.Windl
Version: Leap 42.1Flags: ancor: needinfo? (Ulrich.Windl)
Target Milestone: ---   
Hardware: x86-64   
OS: openSUSE 42.1   
Whiteboard:
Found By: --- Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Attachments: Output of "fdisk -l"

Description Ulrich Windl 2015-12-15 12:34:24 UTC
A newly installed default system does not boot on a HP DL380 G5 Server.
The BIOS just reports a failed boot and repeats the boot forever. Boot device is on a logical partition of a P400 SmartArray controller. However I suspect it's a problem of GRUB2.
Comment 1 Ulrich Windl 2015-12-15 13:38:00 UTC
I think I found the problem: The Array had three logical disks defined, but only the first being scanned for boot code. The installation used the largest logical disk to install everything (it seems). After a new installation with manual disk partitioning, the system booted.
Comment 2 Ancor Gonzalez Sosa 2015-12-16 09:13:48 UTC
Ulrich, if I got it right, it was not Grub or YaST fault then. Grub was installed only in a logical disk and the BIOS didn't consider that disk for booting. Right?

I guess there is no much we can fix about this.
Comment 3 Ulrich Windl 2015-12-16 09:38:13 UTC
As I wasn't patient enough to collect the installation state before doing another installation, it's a bit hard to guess was had happened.
Fact is that there was Windows/2003 on the first logical disk and that booted. So I guess that partition was marked active. As installation wiped the NTFS filesystem, I guess the boot sector on logical drive remained intact, while the new boot loader was written to another logical drive. So naturally when booting the boot sector on the first logical drive had a small chance to succeed.
So the primary error of installation was to use the largest drive instead of the first drive for installing the system boot area.
Comment 4 Ulrich Windl 2015-12-16 09:40:30 UTC
(In reply to my comment #3)
> (...) the boot sector on the first logical drive had a small chance to succeed.

That's the polite version of: "the boot sector had no chance to succeed" ;-)
Comment 5 Ancor Gonzalez Sosa 2015-12-16 09:43:28 UTC
Josef, what's your view here?
Comment 6 Josef Reidinger 2015-12-16 10:20:12 UTC
For me it looks like problem with "boot" flag on disks. I need logs to check where it is set. Choosing the biggest drive should not be problem for bootloader as long as boot flag is properly set and as long as disk boot order in bios is correct (something installer cannot change). But as I said I need logs.
Comment 7 Ulrich Windl 2015-12-16 11:34:59 UTC
Created attachment 659542 [details]
Output of "fdisk -l"

The logical drives look like this:
/dev/cciss/c0d0: 100 GiB
/dev/cciss/c0d0p1: 100 GiB, bootable
/dev/cciss/c0d1: 5 GiB
/dev/cciss/c0d1p1: 5 GiB, NTFS (some recovery partition)
/dev/cciss/c0d2: 174.4 GiB, unpartitioned!

So if /dev/cciss/c0d2 is unpartitioned now, I guess it was unpartitioned all the time. These are the facts I can get from the system as it's running now.
Comment 8 Ancor Gonzalez Sosa 2015-12-16 14:08:30 UTC
Ulrich, Josef was referring to the YaST logs generated during installation.
https://en.opensuse.org/openSUSE:Report_a_YaST_bug
Comment 9 Arvin Schnell 2016-01-14 14:45:51 UTC
Requested information not provided within 4 weeks. Feel free to reopen the bug
once the information is provided.