boot disks progress... and problem :-(
Hello, in trying to implement the diskless installation, I'm running in troubles with the loop device: I cannot mount the rescue & drivers images through the loop service because they reside on an NFS server and the loop device does not allow to read from NFS. Until now, I lived with this bug by coping the image to the target disk before mouting it. I was able to do it because the installation was dedicated to local disks only. However, on diskless workstation, there is no success to copy the image to the target disk since it is also served by NFS :-( Therefore, the rescue & drivers images cannot be used anymore to do such installation. I tried both 2.0.35 & 2.1.125 kernel with no success at all. I don't know why the loop device is broken on NFS images. Any pointers ? Thanks in advance. PS: the kernel in kernel-image-2.1.125 is not compiled with NFSROOT, BOOTP & RARP options as the 2.0.35 did. Theses options are required for diskless workstations & low memory installation (where the root image cannot fit in memory). Christian, can you upload a new package with these options enabled? -- Eric Delaunay | "La guerre justifie l'existence des militaires. [EMAIL PROTECTED] | En les supprimant." Henri Jeanson (1900-1970)
Re: The time has come (part II) ...
> which kernel will we use as standard kernel ? > 2.0.35 is pretty stable for me but it's not completely stable because the > virtual consoles die sometimes. 2.1.125 is much better but it has got the > X logout crash problem. A recent 2.1 kernel would be nice since it supports SMP. Any chance of going to something newer than 2.1.125? Cheers, Anton
Re: The time has come ...
> I updated my cvs snapshot a few hours ago, and there was still a bug. > But applying the patch that you sent to the debian-sparc mailing list fixed > the problem. Thanks! Great. The fix has been checked into the cvs tree. Cheers, Anton
Re: The time has come ...
> Alas :-( It doesn't fix it. The machine stil halts (actually, it watchdog > resets) when X dies. Can you get a dump of .registers, .locals, ctrace? I don't have a cgsix so it is pretty hard to trace without something to go on :) You could set up a serial console on the sun if the screen is stuffed. Cheers, Anton
I've installed the latest sparc32 Debian on an Ultra5.
"Marcelo E. Magallon" <[EMAIL PROTECTED]> writes: > On Wed, Nov 25, 1998 at 03:58:36PM -0500, Steve Dunham wrote: > > > Has anyone tried this stuff on an Ultra? Does it need any dynamic > > loader patches or will an Ultra work out of the box with a special > > kernel. If the latter is true, then we should provide either a > > seperate tftpboot image and floppy set for the Ultra, or a combined > > "TILO" image and seperate floppy set. > I'm going to... today hopefully. I was going to ask if that was possible > but I said "the heck with it, let's give it a try". I fact, I was wondering > why the ultralinux port is based on ultrapenguin and not the Debian Sparc > port. Well, to get you started, go to: http://www.cse.msu.edu/~dunham/debian/sparc/ Get tiloboot.debian It is a TILO tftpboot image with the 2.1.126 kernels from UltraPenguin and the root.bin from Debian. It supports both UltraSparc (sun4u) and sparc32 (sun4[cdm]) machines. base_sid.tgz You need this becase the current base package contains an older libc6 which doesn't play well with 2.1.x kernels. This has all of the packages updated and includes "apt". vmlinux64-2.1.126.gz You need to install this manually (and rerun silo) after the base is installed. modules64.tgz These are the modules for the above kernel. Untar (zcat|star) into /target. modutils64.tgz These are the 64bit modutils. They complain if they are not named /sbin/foo (rather than /sbin/foo64), so you probably want to rename them. Delete /target/sbin/rmmod and untar this in the root of /target. Unfortunately, the is no HME support in the kernel, and the floppy doesn't work on Ultra5 machines, so I added "sunhme.o", "3c59x.o" and "sunqe.o" to the root directory of the ramdisk. You will need to load the appropriate module by hand. (The Ultrapenguin insmod and insmod64 have been installed on the ramdisk.) If we want Ultra support in Debian, we can make "silo.conf" look like: root=/dev/sda1 timeout= 10 image[sun4u]=/boot/vmlinux64-2.1.126 label=linux image[sun4c,sun4d,sun4m]=/boot/vmlinux32-2.1.126 label=linux and it will automatically use the right kernel. So we are really close to Ultra support in the sparc32 version of slink. We would only need to adapt our tftpboot image, add the UltraPenguin version of modutils, add sparc64 kernels, and add the tools for building sparc64 kernels. (We also should change the sparc32 "nm" so it recognises sparc64 - needed for "maketilo".) Steve [EMAIL PROTECTED]
Re: SCSI disk on Sparc IPC
> I'm trying to install linux on a Sun IPC. > > I've put an HP hard disk drive D2077 (2.1 Gb), and when installing > Solaris to format an label the disk, it doesn't recognize it. > > I haven't found anything about it in HP's WEBs. > >Does anyone knows the number of heads, cylinders, etc to define it ? > Do I have any other solution ? Hmm, Solaris' format command is usually able to read the geometry from the disk (using SCSI the inquiry command). When you are at format's main menu (after saying which disk you want to format and telling it you are not ready to label it), type "type" and select 0 (auto configure) and it should figure out what you have. You can try this by booting the Solaris cd in single user mode (boot cdrom -s or b sd(0,6,0) -s depending on your what proms you have). It will drop you to a root prompt where you can run format. Regards, /Anders -- -- Of course I'm crazy, but that doesn't mean I'm wrong. Anders Hammarquist | [EMAIL PROTECTED] Not system administrator at DjungelData | Hem: +46 31 47 69 27 Chalmers University of Technology, G|teborg, Sweden | Mob: +46 707 27 86 87
Re: The time has come (part II) ...
"Jules Bean" <[EMAIL PROTECTED]> writes: > --On Wed, Nov 25, 1998 10:08 pm +0100 "Eric Delaunay" > <[EMAIL PROTECTED]> wrote: > > > About kernel, I will release the next bootdisks with 2.0.35. Is it enough > > reliable for *all* sparc workstations, or should I switch to 2.1 ? I'd like to see 2.1, but I can always make my own... > Since the last breaks dpkg-shlibdeps, I'm using 2.1 ATM. But it does have > some rather serious X problems on cgsixes (?) Do you have a flashing cursor in X with the cgsix? (I do, with 2.1.127, and I'm trying to determine if it's due to the fact that the cgsix is the second head on my system.) Steve [EMAIL PROTECTED]
Re: Problem with sendmail 8.9.1-4 or libc6 2.0.100-2.1?
"J. S. Connell" <[EMAIL PROTECTED]> writes: > When I attempt to run any of the binaries included with the sendmail > 8.9.1-4 package, I receive the following error: > /usr/sbin/sendmail: error in loading shared libraries: /usr/sbin/sendmail: > undefined symbol: db_open > ldd reports: > > libdb.so.3 => /lib/libdb.so.3 (0x5002c000) > libresolv.so.2 => /lib/libresolv.so.2 (0x50084000) > libnsl.so.1 => /lib/libnsl.so.1 (0x500a6000) > libc.so.6 => /lib/libc.so.6 (0x500d1000) > /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x5000) > > I have the latest versions of these libraries (as of about 11 am this > morning) installed. > I can't check libdb.so.3 for db_open because it doesn't have symbols (so > says nm). # nd -D /lib/libdb.so.3 000406d8 T __db_ndbm_open 00040be0 T __dbopen 000163f0 T __nss_db_open 000163f0 W db_open 000406d8 W dbm_open 00040be0 W dbopen U fopen 00034b40 T lock_open 00036270 T log_open 0003b560 T memp_fopen 0003c900 T memp_open U open U opendir 0003e7b0 T txn_open # It seems to have the symbol. And sendmail loads for me. sendmail 8.9.1-4 libc 2.0.100-2.1 Steve [EMAIL PROTECTED]
SCSI disk on Sparc IPC
I'm trying to install linux on a Sun IPC. I've put an HP hard disk drive D2077 (2.1 Gb), and when installing Solaris to format an label the disk, it doesn't recognize it. I haven't found anything about it in HP's WEBs. Does anyone knows the number of heads, cylinders, etc to define it ? Do I have any other solution ? Thanks.
Re: ash need to be recompiled
On Fri, Nov 27, 1998 at 11:47:38PM +0100, Eric Delaunay wrote: > Christian Meder wrote: > > Very cool. Would it be possible to create a 2.0.35 installation set > > and a 2.1.130 set (next week) ? best of both worlds approach I guess > > ;-) > > I will try to upload new boot disks next monday, based at least on 2.0.35. > I don't know if I will have time to work on 2.1 series. The serial console > code detection have to be reworked since 2.1 now use a special device for the > console (no need to install the right symlink to the tty). > Nevertheless, if I'd do it, it will be based on 2.1.125, not 2.1.130. This is > the only one present in my local mirror (I don't have regular access to a fast > internet connection, just a 14kb/s modem every days & a 2MB/s one or two times > per week when I go to my lab ;-( ). But I promise you I will get 2.1.130, or > whatever the number is, as soon as possible then repackage the boot disks. > > Regards. > > PS: I'm also working on diskless installation for those interrested in. Thank you very much for all your efforts. Christian -- Christian Meder, email: [EMAIL PROTECTED] What's the railroad to me ? I never go to see Where it ends. It fills a few hollows, And makes banks for the swallows, It sets the sand a-blowing, And the blackberries a-growing. (Henry David Thoreau)
