I think a NBD device is just like any other block device, isn't it? I was expecting it to have things like ID_FS_TYPE set on it.
The nbd devices are set up loop devices, we first have dead device
nodes, then later we attach stuff behind them. At the time of the udev
creation event, nothing works at nbd devices.
Userspace has no idea when to probe for the device, or when stuff changes,
it needs proper kernel support for that to work, which is as far as I know
It is currently blacklisted for that reason, and removing the blacklist
could only happen if the nbd kernel side is changed to send out events
like loop, dm, md are doing it.
(In reply to comment #1)
> The nbd devices are set up loop devices, we first have dead device
> nodes, then later we attach stuff behind them. At the time of the udev
> creation event, nothing works at nbd devices.
That, unfortunately, makes a lot of sense. Thanks a lot for the explanation!
It should be possible though, to fix that in the kernel. It should be
just a couple of lines, with uevent sending hooks at the right place.
People familiar with nbd should be able to do that pretty quickly.