F16 Alpha GRUB install failure

[Chris Murphy]

On Feb 13, 2012, at 7:10 PM, Adam Williamson wrote:

> On Mon, 2012-02-13 at 14:25 -0700, Chris Murphy wrote:
>> 
>> I think the problem is GRUB2's own install script/app, doesn't do a great job of accounting for disks partitioned where the 1st partition comes less than 35KB after the MBR, and as the core.img is too large it fails to install between the MBR and partition 1.
>> 
>> Strangely though, anaconda manages to get it to install without a complaint.
> 
> No, that's clearly not the problem here, because this thread is about
> installing grub to the front of a partition - *not* to the MBR.

The GRUB2 manually doesn't outright say "unsupported" or "not recommended" but it does say insertion prior to the 1st partition is what's recommended. I've had zero problems with GRUB legacy installing into specific partitions, but worse than 50% failure with GRUB2 so I've given up on stuffing it into specific partitions.


> anaconda only calls grub2-install, with appropriate parameters, to
> install grub. It doesn't do anything particularly special or clever.

Well I've got a number of cases on F16 where anaconda's call to grub2-install produces a smaller core.img than calling grub2-install directly. I don't have an explanation for this, but it's just enough of a difference in size that it causes problems with older partitioning schemes that start partition 1 at sector 63.

Is anaconda passing --force by default?


Chris Murphy