Date: Fri, 14 Jul 2000 09:27:58 -0600 (MDT) From: Fred Clift <fred@veriohosting.com> To: John Baldwin <jhb@FreeBSD.ORG> Cc: Greg Lehey <grog@lemis.com>, "Anders Chr. Skoe" <skoe@owlnet.rice.edu>, freebsd-questions@FreeBSD.ORG Subject: Re: Anyone resolved "Missing operating system" issue? Message-ID: <Pine.BSF.4.21.0007140840410.37149-100000@vespa.orem.iserver.com> In-Reply-To: <XFMail.000713210002.jhb@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> > I've never seen any problems using "dangerously dedicated" on SCSI. > > I've been doing it all the time for over 8 years. Care to explain > > your claim? > > I know that Doug White for one had serious issues with dangerously > dedicated mode on SCSI. The problem has more to do with newer SCSI > controllers that attempt to match the geometry speecified in the MBR > of the disk. Since dangerously dedicated mode uses a completely fake > and made up geometry, this can greatly confuse the SCSI controller. Intel ISP2150 servers (2U case, with the L440GX motherboard+hotswap stuff) have an adaptec scsi bios built in to the mother board that kind of chokes on dangerously dedicated disks. At the first reboot, the box hangs. In fact, if the 'fake' freebsd fdisk partition talbe that gets put on dangerously dedicated disks (the one with the 50 meg entry) makes the machine choke if it is on _ANY_ disk in the system, even if other devices are supposed to be the boot devices (as chosen by bios setting). The bios sees what purpots to be a valid partition table on sector 0 (last two bytes AA 55 or 55 AA) and then finds an entry in the fourth partition slot marked active with enoguh info to make most motherboards boot the disk fine, but with enough wrong info that the machine totally chokes. I assume that it is trying to verify that there is some valid filesystem in the partition or something. I did find a workaround for this particular board -- I do an fdisk -u on the drive after it has been all installed properly. I go to the partition entry that has the bogus 50 meg freebsd partition and then interactively change the value for 'size' to be the same size that disklabel shows and then let it automatically calculate the beginning/ending addresses and write the partition back. (note that I tried to use the scripting stuff built into the fdisk command, but getting the same numbers is virtually impossible because the scripting stuff thinks it knows better than you do and is in my opinion broken. I end up sending something like "\n\n\n\ny\n\n\n%d\n\ny\n\ny\n" ,driveInfo[disk].size to the fdisk -u process which gets the job done but is ugly. One of these days, I'll fix fdisk instead and submit a patch :) Then I have a 'dangerously dedicated' but have a 'real' fdisk entry for the partition. This fools the so-called 'smart bios' into actually booting the disks... Of course this could be avoided alltogether by not doing dangerously dedicated disks, but... Fred -- Fred Clift - fred@veriohosting.com -- Remember: If brute force doesn't work, you're just not using enough. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0007140840410.37149-100000>