Re: New VM image

["Ludovic Courtès" <ludo@gnu.org>]

Ludovic Courtès <ludo@gnu.org> skribis:

> Chris Marusich <cmmarusich@gmail.com> skribis:

[...]

>>> (define this-file
>>>   (local-file "./vm-live.tmpl" "config.scm"))
>>
>> Cute!  But FYI, I tried building this and it failed at first, since I
>> renamed the file and forgot about this reference.  I then re-ran the
>> command from the same directory as the file, which of course worked.
>
> Yeah, we could maybe extract the actual file name from
> (current-source-location).

Done in 1ac6c33caec2c2a4aaadda49cb0febb5bfbf264c.

>> When I booted into the system, I noticed some things:
>>
>> - There are many folders on the desktop.  For example, there is one for
>>   "/sys/fs/cgroup/memory".  A minor blemish, but rather odd.
>
> Yeah, that’s a longstanding issue with GLib/GIO: that library has code
> to find out which mount points are “system” mount points (things that
> should not be shown in the UI), but for some reason, it doesn’t work
> correctly for us:
>
>   https://lists.gnu.org/archive/html/help-guix/2017-11/msg00087.html

Fixed in 74685a4369033e79a776c5dec1c8480e8b446f6b.  The solution turned
out to be disappointingly trivial and is another example of this Unix
baggage we’re still carrying…

>> - "sudo herd status" reports that term-auto and user-homes is stopped,
>>   which I think might be intended, but I'm not sure.  I also see
>>   possibly unnecessary services: wpa-supplicant, upower-daemon.  The VM
>>   probably doesn't need wpa-supplicant, but maybe it needs
>>   upower-daemon?  Not sure.
>
> True, the VM currently uses NM + wpa_supplicant, which we could
> certainly replace with the DHCP client.

Done 

>>   I then ran "guix pull"
>>   and tried again with "sudo -E guix system reconfigure
>>   /etc/config.scm", since I remembered that the proper way to do this
>>   using my freshly pulled Guix would be to use -E.  I got a new error:
>>   "error: device '/dev/vda' not found: No such file or directory", which
>>   occurred because the VM has /dev/sda but not /dev/vda.  Maybe we can
>>   modify the config file to use sda instead?
>
> It all depends on how you start QEMU: the device would be /dev/vda if
> you use virtio, and /dev/hda otherwise.  In the manual, we should
> probably give the incantation that leads it to use virtio:
>
>   -device virtio-blk,drive=myhd -drive if=none,file=/tmp/t.img,id=myhd

I’ve done that in 05b60752a67af294c35d8f4218487c9a44b5fdee.

I’ve also fixed sudo along the lines of what Danny suggested in commit
6e0521eeede6bd06bc083073145413a04771aad3.

I think it’s pretty decent now, or at least there are fewer rough edges!
:-)

Ludo’.