Date: Sun, 19 Mar 2006 17:58:56 -0600 From: Martin Jackson <mhjacks@swbell.net> To: freebsd-stable@freebsd.org Subject: Re: 6.1-BETA4 installation report Message-ID: <441DF040.7040704@swbell.net> In-Reply-To: <441DE846.1070309@errno.com> References: <441DC164.5060703@swbell.net> <20060319215507.GA7464@xor.obsecurity.org> <441DDEA6.8090805@swbell.net> <441DE846.1070309@errno.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Sam Leffler wrote: > Martin Jackson wrote: >>>> 1) The upgrade errored out almost immediately on /var/empty, since >>>> it had been chflags'ed to nschg. I went to the shell and chflags'ed >>>> that off, and upgrade continued. >>> >>> It's supposed to be that way, so something is wrong if this is causing >>> problems :( >> >> Agreed. :) Looking through the source to sysinstall, it looks like I >> should have a /boot/kernel.prev, but I don't. Perhaps the install of >> base went more poorly than I thought? I wound up not having >> /boot/kernel at all. > > If you go to the Options panel and turn on debugging (navigate to > Debugging, hit space, then q to quit) you'll get debug info in the > Alt-F2 window. That might be helpful. I'm not sure if the logging goes > in a file that remains across reboot; that'd be useful for debugging > issues like this. > >> >>>> 2) The 'kernels' distribution wasn't installed at all, which led >>>> loader to complain upon reboot. I had an old kernel installed, so I >>>> booted from that and used sysinstall to install GENERIC. I also had >>>> to change loader.conf to reference 'kernel="GENERIC"'. Is this the >>>> preferred way to do it, or should I move 'GENERIC' to 'kernel'? >>> >>> It should be installed as kernel. Again I don't know how you got >>> something else :( >> >> Hm. I will try to reinstall with 6.0-RELEASE and try to do the binary >> upgrade again, and see what happens. Things will be different; this >> system was my first attempt to go to STABLE before I upgraded it so I >> can't rule out errors of my own in this case. > > If you can record the steps by which you hit a problem it'd be > appreciated. I'm not sure I tested the Upgrade mechanism so it may have > been broken by the mods I made to the kernel install logic. OK, I've hit the /var/empty problem again, on a fresh install of 6.0-RELEASE. I enabled debugging but it doesn't say anything else useful on the F2 window. Installing base, kernel (GENERIC), doc (custom set) from CD/DVD Here's what it says on the F1 window (Transcribing): Write failure on Transfer! (wrote -1 bytes of 1424508 bytes) On the F2 window: ...list of the fs hierarchy... var/crash var/cron var/db /stand/cpio: var/empty: chmod: Operation not permitted /stand/gunzip: failed fwrite I go back to F1 and accept. When it asks me to retry, I go to F4 and chglags noschg /var/empty. I go back to F1 and say "YES" to attempt to retrive again. This time, it seems to go correctly. I see that GENERIC is trying to extract into /boot this time - don't remember whether it said that the first time around but I see it as a positive sign. /boot/kernel exists this time (no special intervention from me), and contains the GENERIC kernel from 6.1-BETA4 CD. No special intervention required to boot this time; system is now up and running on 6.1-BETA4. Well, it looks like unless there's something going on with custom ident's in /boot/kernel, that may be a non-issue. (The previous kernel was ident SERVER.) The /var/empty thing looks to be real, though. Should I file a PR? Thanks, Marty
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?441DF040.7040704>