Re: rootfs goes full during installation of 5.0.1 on NWS-1460

[Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>]

> Hello, is anybody here? ;-)

Yes, of course :-)

> I have a NWS-1460 with 16 megabytes of memory, which I'd like to
> install NetBSD on.

Wow, I'm really amazed to see still a new user of this port,
even with a working ~20 years old NEWS machine ;-)

> I downloaded 5.0.1 boot.fs from one of the mirror,
> booted it, ran the installer, and after the message
>  > I found only one disk, sd0.
>  > Therefore I assume you want to install NetBSD on it.
> the installer terminates with following output:
>  > uid 0, pid 7, command sysinst, on /: file system full
>  >
>  > pid 7 (sysinst): user write of 28672@0x12c000 at 1112 failed: 28
>  > /: write failed, file system is full
>  > [1]   Illegal instruction     /sysinst
>  > #

Hmm, on my NWS-1750, 5.0.1 sysinst works without problem.
(at least it goes though newfs root file system)

On 5.0.1 installation disk, ramdisk size is actually tight
so something might flood ramdisk root easily.
By default, df(1) should print as the following right after boot:
---
# df
Filesystem  512-blocks       Used      Avail %Cap Mounted on
/dev/md0a          2767       2729         38  98% /
#
---

Is there any file in /tmp (or other directories) after crash?
Did you enable logging in the utility menu, for example?
If so, "mount -t mfs swap /tmp" command before starting sysinst
might work around.

> I tried copying all the contents of rootfs to one of my NFS servers
> and chrooted into it from the INSTALL kernel from the floppy,
> then run /sysinst manually, but no luck: the installer terminates with
> bus error.
> 
> Does anyone here have the same problem? Is there a solution?

I'm not sure which point the bus error occured on,
but another possibility is that the NetBSD kernel was
confused by existing old disklabel on your sd0 disk.
What "disklabel sd0" on the ramdisk prompt shows?

---
Izumi Tsutsui