Date: Mon, 09 Oct 95 00:27:19 PDT From: Ben Jackson <bjj@sequent.com> To: bugs@freebsd.org Subject: install of 2.1.0-951005-SNAP Message-ID: <199510090727.AAA13577@crg8.sequent.com>
next in thread | raw e-mail | index | archive | help
Someone recently gave me an Archive Viper 150, and I had to be at work on Sunday, so it was the perfect opportunity to upgrade and get rid of the pesky ahc scsi bus lockups that made it impossible to back up /home. First, let me say (to Jordan?) that the install looks really snazzy. I was impressed by all of the new features. I did run into a few bumps, though: 1. When the boot floppy kernel boots, after uncompressing, it says it is booting the "kernelel" [sic]. 2. after I selected my install media (UFS), the shell escape to mount it didn't appear. I searched high and low for a way to get a shell (and even noted that `sh' was installed in /stand). I finally found it on VT3. I actually tried VT2 early in my search, but it didn't have anything, so I had abandoned that line of inquiry. this was really only necessary because I had the path wrong (after disklabeling and commiting, it worked). 2a. stty settings on root shell were wrong (eg ^? vs ^H) 3. re: (2), I was panicked momemtarily after my root partition had been newfs'd, and I couldn't get it to install the tarballs. It might be clever to check and make sure that at least bin.aa is accessible before wiping out the previous installation. 4. ...which reminds me, the 'upgrade from 2.0.5' was a cruel trick! I was disappointed to find it a nop. As far as I can tell, now that I'm done, if I'd just been able to evade that newfs, and a few key config files had been spared (or converted), it would have worked. sparing config files could be as simple as a 'initial config' dist that is conditionally extracted. 5. many dialogs of the "oops, info, we're continuing" have `<Exit>' at the bottom. The first time I got one I spent some time trying to figure out how to avoid it, since in previous installs, selecting `<Exit>' always led to starting over from scratch. I think `<Continue>' would be more intuitive. 6. when it couldn't find my install media, after asking if it should continue, it did regardless of my answer. 7. even dists which fail to install are deselected after an extract pass. 8. in the disklabel menu, the second column of partitions is messed up. the areas that were hilited for me, and the number in the select sequence: 1 2 3 4 12 8 5 9 6 10 7 11 so if you moved from '7', it hilited the end of the first entry in the second column (which started as '*WAP', like it had drawn some bogus swap entry first). 12 wrapped to 1 and vice versa. If I edited the info I could see (which was in the obvious sequence in the table) while I was at `12', it modified the right thing. This made me pretty nervous that a commit would newfs something I didn't want it to. 9. disklabel zapped my bootblocks (which were custom to make the default boot device sd0 even tho it got the block from wd0). 10. timezone setting was kind of neat, but the most common options (ie NA/US) were buried due to alphabetical order. after I gave the current time and selected everything, it presented me with a time that was one hour off (and several years, but I don't know what that was about). It thought I had CMOS clock set at wall time. now that my interactive restore of / has put /etc/localtime link back where I had it under 2.0.5, things are back to normal. 11. after first reboot, there is no /kernel, only /kernel.GENERIC In general, as the GUI gets gooeyer and gooeyer, I dissolve more and more of my stomach lining during an install (I just know one day that newfs is going to go nuts and hit partitions I want to keep...) BTW, all of this was via "custom" install. I didn't see any of the quick install or much of the system configure (tz botch convinced me to just do it by hand). Thanks all, --Ben
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199510090727.AAA13577>