Prev: Solaris 10 - change IP address
Next: pcf8584: i2c_0 lost arbitration
From: Logan Shaw on 13 Nov 2005 00:31
I'm getting the message in the Subject: line when I try to
install Solaris 10 GA on an x86 system (well, actually amd64).
I'm trying to do a set of slices that looks like this:

0 / 10GB
1 swap 10GB
7 /export/0 95GB

And, the thing is, I agree that it does extend beyond cylinder
1023, but I thought this wasn't a problem with modern BIOSes.
And the machine I have is fairly new -- it's an nforce3 250
chipset, an MSI K8N Neo motherboard to be specific. Shouldn't
a board made within the last year or so be able to handle booting
off a partition that extends past cylinder 1023?

For what it's worth, I have checked the settings for the boot
drive in the BIOS, and there are 4 choices: CHS, LBA, Large,
and Auto. I have it set to Auto, but since this is a 120GB
drive, I assume it MUST be picking either LBA or Large[1]
since CHS wouldn't be large enough to handle this 120GB drive.
I've also tried it with it explicity set to LBA and it doesn't
seem to make any difference.

So, my questions are:

1. Is this limitation really still a problem, or is the
installer just being picky so that for the (older)
systems on which it IS a problem, people won't run
into trouble?

2. If it's not really a problem for a newer BIOS, is there
a way to tell Solaris that yes, I really do know that
going past 1023 is OK? I could make a smaller root
filesystem, but I'd rather not have to have a separate
/usr and /var, which is what that would require.

3. Would an x86boot partition (that resides entirely in
the range 1-1023) help? I tried doing one and it
didn't seem to make a difference, but maybe I missed
something. Also, I think I remember hearing that
x86boot was going the way of the dodo at some point.

Any suggestions?

- Logan

[1] whatever "Large" means -- LBA48, perhaps?
From: Casper H.S. Dik on 13 Nov 2005 10:10
Logan Shaw <lshaw-usenet(a)austin.rr.com> writes:

>So, my questions are:

>1. Is this limitation really still a problem, or is the
> installer just being picky so that for the (older)
> systems on which it IS a problem, people won't run
> into trouble?

No; it's not a problem but under certain circumstances the
error crops up.

>3. Would an x86boot partition (that resides entirely in
> the range 1-1023) help? I tried doing one and it
> didn't seem to make a difference, but maybe I missed
> something. Also, I think I remember hearing that
> x86boot was going the way of the dodo at some point.

No; don't do a x86 boot partition.

Casper
From: Logan Shaw on 13 Nov 2005 18:16
Casper H.S. Dik wrote:
> Logan Shaw <lshaw-usenet(a)austin.rr.com> writes:

>>So, my questions are:

>>1. Is this limitation really still a problem, or is the
>> installer just being picky so that for the (older)
>> systems on which it IS a problem, people won't run
>> into trouble?

> No; it's not a problem but under certain circumstances the
> error crops up.

Hmm... So is there any way to tell the installer that I
don't care about the 1023 cylinder boundary and to go ahead
and do the install anyway? Right now it's a fatal error.
The installer pops up a window with the message, and when
I dismiss the window, the installer exits. (It doesn't even
have a "back" option where I could change the partitioning!)

- Logan
From: Casper H.S. Dik on 14 Nov 2005 10:30
Logan Shaw <lshaw-usenet(a)austin.rr.com> writes:

>Casper H.S. Dik wrote:
>> Logan Shaw <lshaw-usenet(a)austin.rr.com> writes:

>>>So, my questions are:

>>>1. Is this limitation really still a problem, or is the
>>> installer just being picky so that for the (older)
>>> systems on which it IS a problem, people won't run
>>> into trouble?

>> No; it's not a problem but under certain circumstances the
>> error crops up.

>Hmm... So is there any way to tell the installer that I
>don't care about the 1023 cylinder boundary and to go ahead
>and do the install anyway? Right now it's a fatal error.
>The installer pops up a window with the message, and when
>I dismiss the window, the installer exits. (It doesn't even
>have a "back" option where I could change the partitioning!)


Run fdisk; add the Solaris partition; restart the install.

That seemed to fix it for me; I think it may read a bad disklavel
and then complain.

Casper
--
Expressed in this posting are my opinions. They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.
From: Logan Shaw on 15 Nov 2005 04:48
Casper H.S. Dik wrote:
> Logan Shaw <lshaw-usenet(a)austin.rr.com> writes:
>>Casper H.S. Dik wrote:
>>>Logan Shaw <lshaw-usenet(a)austin.rr.com> writes:

>>>>1. Is this limitation really still a problem, or is the
>>>> installer just being picky so that for the (older)
>>>> systems on which it IS a problem, people won't run
>>>> into trouble?

>>>No; it's not a problem

>>is there any way to tell the installer that I
>>don't care about the 1023 cylinder boundary and to go ahead
>>and do the install anyway? Right now it's a fatal error.

> Run fdisk; add the Solaris partition; restart the install.

Thanks, that seems to have done the trick.

> That seemed to fix it for me; I think it may read a bad disklavel
> and then complain.

Or perhaps it complains only if *it* is creating a fdisk
partition table, but if the fdisk partition table already
exists, then the creating/complaining code is bypassed.

- Logan
 | 
Pages: 1
Prev: Solaris 10 - change IP address
Next: pcf8584: i2c_0 lost arbitration