Re: grub fails to boot, prints only a single dot

[Andrei Borzenkov]

On Tue, Jun 9, 2015 at 6:08 PM, Alex Regan <address@hidden> wrote:
> Hi,
>
>>> You mean after running it with the debug options you provided? Is that
>>> really going to make a difference? Or should I see something else printed
>>> during boot with these options?
>>
>>
>> I hoped so (as I said --debug-image should enable quite verbose output on
>> boot).
>
>
> How do I capture that output?
>

I'm afraid only by making a photo of screen where it stops.
...

>
>> Unfortunately I do not really see
>> anything wrong so far. The worst case would be your BIOS having issues
>> with more that 2 disks; because this is the only explanation I see so
>> far. GRUB needs to scan all disks to find MD RAID.
>
>
> Ah, interesting. I thought I had other similar systems with four disks, but
> perhaps not. This could also mean that reinstalling still doesn't get it to
> work, because I would have /boot on RAID1 then too.
>

For a test you could split /boot and leave it on physical disk just to
check whether GRUB can boot in this case (I expect it can). This could
give you workaround - you still can manually copy kernels and updatre
grub on second disk.

> Could it have to do with the version of grub? It's a fedora20 system, so not
> all _that_ old.
>

it is always possibility; but as I understand you are not in position
to easily experiment with different versions. Although testing
upstream GRUB is always helpful.

> Does it help to know that it's a SuperMicro X7DBP motherboard? I believe
> it's this chassis with four INTEL SSDSC2CW060A3 SSDs:
>
> http://www.supermicro.com/products/system/1U/6015/SYS-6015P-TR.cfm
>

No, sorry. No experience with particular hardware.

> Is it also curious that the sysrescue CDROM is able to boot to the system on
> the disks without any trouble?
>

Well, it just reads from CD and then Linux kernel takes over. My
suspect was BIOS where proram attempts to scan for/access drives.

BTW do you have /sys/firmware/edd directory (from memory, could be
different) or something related to EDD in dmesg log?