Port-mac68k archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: NetBooting a Quadra from MacOS X




Am 29. Apr .2008 um 21:37 Uhr schrieb Hauke Fath:

tset(1), touch(1) are from /usr/bin, df(1) is from /bin. It would seem that
you accidentally picked a usr tarball for a different architecture?

I don't know what was wrong here. I simply recreated the whole /export/macbsd/root thing from scratch (downloading base.tgz, comp.tgz, man.tgz, and etc.tgz from ftp.netbsd.org/pub/NetBSD/NetBSD-4.0/mac68k/binary/sets) and used tar -xvpzf /export/macbsd/NetBSD-4.0/binary/sets/xxx.tgz on each of these. I cannot recognize any difference to what I was doing before, but now the binaries in "/usr/bin" are usable :-)

About shoehorning a MacOS X machine into being a proper nfs server I cannot
say much

I have Leopard (10.5) installed on my iBook at home. I just tested to set up NFS by creating an /etc/exports file with a "/export/macbsd -alldirs" line in it, and it worked instantly without any complaints (as I said, there is no more NetInfo and no more mountd in Leopard). So it seems that something is broken in Panther (10.3). Unfortunately I cannot upgrade the Mac used here for several reasons, but I won't bother any longer. Right now I'm only exporting "/export/macbsd/root -maproot=root:wheel 192.168.101.69". To solve the swap-problem, I created a "swap0" file in a "/swap" directory of the client. Then I tried successfully

        # swapctl -a  /swap/swap0

My fstab now looks like:

        /swap/swap0                             none    swap    sw 0 0
        bluex:/export/macbsd/root       /               nfs             rw 0 0

I had to write it that way (sort of pretending it would be a local device?) instead of bluex:/export/macbsd/root/swap/swap0 (where it really resides). Apparently, it is working, even after a reboot:

        swapctl: adding /swap/swap0 as swap device at priority 0

Surely, this is not the best/fastest solution, but I don't expect the Quadra to use virtual memory too often, anyway. Having solved this issue, I have new problems now. When I recreated the clients directories and files, I forgot to do a:

        mknod /export/macbsd/root/dev/console c 0 0

as described int the netboot howto. Thus I got a "macbsd /netbsd: warning: no /dev/console" message on the screen. Nonetheless everything was mostly working (single and multiuser boot) until I "fixed" it by using the above mknod-command. Now the multuser-booting hangs after starting cron. But before that happens, I see a couple of other problems. One is that I have no dump-device configured:

        Checking for core dump...
        savecore: /netbsd: kvm_openfiles: /netbsd: No such file or directory

As I understand it, I would need a dump device for debugging purposes, and usually (part of) the swap device is used for that, correct? However, my swap is only mounted by swapctl after savecore spits out its message. What shall I do about it, just ignore that? Where would the /netbsd directory come from? I created one, but then it says

        savecore: /netbsd: kvm_nlist: bad namelist
        savecore: /netbsd: _dumpdev not in namelist

That would have been too easy, obviously...

Andreas



Home | Main Index | Thread Index | Old Index