Re: 9.0 amd64 install assertion

[Steffen Nurpmeso <steffen%sdaoden.eu@localhost>]

Hallo Martin.

Sorry for the late response.

Martin Husemann wrote in
<20200402060344.GA2400%mail.duskware.de@localhost>:
 |Please describe the state of the disk before you started (empty, an old 
 |installation with disklabel/MBR, ....) and how exactly you got to the
 |point of failure (i.e. all menu selections).

Just like i said, everything default nothing special.  qemu with
8 GB disk, looked in custom partitioning but choose the NetBSD
provided defaults, used default sizes etc.  This means two
partitions, one for / one for swap i think.  So but then i just
quit instead of installing.  After four reboots, two from harddisk
which booted but stopped because getty could not attach to
/dev/console (one ISO boot in between to run MAKEDEV all which did
not help) i gave up and headed towards installation via HTTP.

So sure, let's repeat it.

  │>a: Installation messages in English
keyboard:
  >a: unchanged
  >a: Install NetBSD to hard disk
continue:
  >b: Yes
Available disks

  │>a: wd0 (8.0G)                                     │
  │ b: 6161d776-d99d-48fc-8c4f-f1534c64ffda (dk0@wd0) │
  │ c: 4a12784d-6024-42a7-8f15-c50a4236933c (dk1@wd0) │
  │ d: Extended partitioning                          │
  │ x: Exit

  - Choose b:
    >b: Use default partition sizes
  assertion

I mean, i do not want that (b).  That is, i wanted to install the
sets to (b), yes.  b and c are of course the result of the initial
install that i aborted in the HTTP menu.
Apropos menu.  The yes/no switches are often very far below the
actual question.

  >c: Re-install sets or install additional sets

 |Also helpfull would be the output of fdisk, disklabel and "gpt show"
 |for that disk (though it now will show the "after" state, not the one
 |that made you run into this issue).

  # gpt show /dev/dk0
  GPT not found, displaying data from MBR.
  start      size  index  contents
  0         1         MBR
  1  12713916         Unused

All others give ioctl errors.

Works nice otherwise, though would be nicer if (tar -cpf - | tar
-xpf -) would save some download and count as installed.  I also
struggled because the e1000 was enabled=0 and unusable, dmesg
however said "are you an emulator?" and so we are now
virtio-net-pci based, which works very good.  However, with qemu
4.2.0 i need to ping the VM before the network works.  I had this
with archlinux 2019.12, even worse, but there a kernel update
fixed the behaviour.  (That was qemu 4.0.0 by then i think.)  I do
not see the behaviour with other BSDs nor Linux, so i thought
i mention it.

--steffen
|
|Der Kragenbaer,                The moon bear,
|der holt sich munter           he cheerfully and one by one
|einen nach dem anderen runter  wa.ks himself off
|(By Robert Gernhardt)