This change removes the bespoke logic around identifying block devices.
Instead of trying to find the right device by iterating over
`qemu.drives` and guessing the right partition number (e.g.
/dev/vda{1,2}), devices are now identified by persistent names provided
by udev in /dev/disk/by-*.
Before this change, the root device was formatted on demand in the
initrd. However, this makes it impossible to use filesystem identifiers
to identify devices. Now, the formatting step is performed before the VM
is started. Because some tests, however, rely on this behaviour, a
utility function to replace this behaviour in added in
/nixos/tests/common/auto-format-root-device.nix.
Devices that contain neither a partition table nor a filesystem are
identified by their hardware serial number which is injecetd via QEMU
(and is thus persistent and predictable). PCI paths are not a reliably
way to identify devices because their availability and numbering depends
on the QEMU machine type.
This change makes the module more robust against changes in QEMU and the
kernel (non-persistent device naming) and by decoupling abstractions
(i.e. rootDevice, bootPartition, and bootLoaderDevice) enables further
improvement down the line.
Test for presence of all specified options in the generated .nspawn
config file.
Additionally test for absence of misspelled and fixed option MachineID.
This is necessary because this test relies on switching the root fs to an empty one which
does not have a Nix store available in stage 1, therefore, we have to make this test
host-store only.
A better fix in the long term is to evaluate whether this is worth to enable a proper
Nix store image for it with EROFS?
I'm merging this without review, since the tests run by ofborg are
succeeding. In addition to that, it's fixing a currently broken test so
the worst that could happen is that the test still does not work.
The test fails because the way the configuration switch was implemented
back then was by using a dummy configuration and simply activating that
dummy configuration from within the test script.
Nowadays, this doesn't work anymore and fails to typecheck because the
dummy "newServer" will inherit the same value for networking.hostName,
which in turn will generate two attributes for "server":
> testScriptWithTypes:43: error: Name "server" already defined on line 43
> [no-redef]
> client1: Machine; client2: Machine; server: Machine; server: Machine;
Fortunately, we don't need to do workarounds like this anymore and there
is the "specialisation" option, which allows to do this in a less ugly
way (and it also works with mypy).
Signed-off-by: aszlig <aszlig@nix.build>
