With such old systems and doing upgrades, there are a few things to worry about compared to updating one version at a time: 1) boot blocks. At least at some point for sparc, my memory is fuzzy, but I had a problem that seems like: I think around NetBSD 3, the boot blocks for NetBSD 3 would not boot 5 kernels. So if you updated to 4 and then installed 4 bootblocks, you were ok. 2) binary compat. NetBSD is really good about binary compat, but I am not sure that the compat options are turned on by default. So booting a netbsd 7 kernel on a 2.0 system may not entirely work. I would recommend one of several options: 0) First, make backups. 1) as coypu@ said, you can put the disk in some other box, upgrade, and put it back. 2) A) Create a script /.installboot to: cp /usr/mdec/boot to /boot, and also to run installboot. Mine is attached, but note that you need to get the root fs type right, do the right thing for raid, and set any serial console options. Run this on the system as it is. B) install sysutils/etcmanage from pkgsrc, and read the etcmanage howto. Register the files that you haven't changed from the system via etcmanage --import, using a manifest that you create from the 2.0 or 3.0 install sets from the archives. C) Get the next version of the OS, and use INSTALL-NetBSD installkernel (from the etcmanage package), and reboot. If that is ok do installuser. Check the etcmanage situation (it will autoupgrade what it can). Repeat for the next version. Optionally, skip versions. As long as the new kernel boots, and you can run the userland install (which involves perl to run etcmanage), you should be able to skip. Probably you can jump from 2 to 4. Probably even to 5. 5 to 7 is probably ok. I have no idea about 2 to 7, but if you just boot a 7 kernel and see if the system is ok, it's likely ok to upgrade if it comes up. -------------------- #!/bin/sh installboot -v /dev/rwd0a /usr/mdec/bootxx_ffsv1 cp -p /usr/mdec/boot /
Attachment:
signature.asc
Description: PGP signature