[ Cc:'ed to the nbd-client bug since things might be quite the same there ]
Christoph Biedl wrote...
> > Also, the SysV init script used to unmount these
> > filesystems, but the aoetools package masks the systemd service. It
> > would be better to provide a real systemd service script that restores
> > the expected behaviour.
>
> Given the fact this is something for stretch and we're pretty far into
> the freeze, there isn't the time for a few loops until such a unit works
> as expected. Therefore I'm asking for help with this topic. I can do the
> testing, then.
After some testing it seems the issue has much less impact than it
seemed initially:
(At least) AoE devices are handled properly if mounted with the _netdev
mount option. This applies to swap-over-AoE, too. Also, mounting these
through /etc/fstab works in jessie only if _netdev is provided as a
mount option as well.
Then the issue boils down to manually mounted AoE devices that will no
longer be umounted early enough. That's something that can be addressed
in the release notes as well. I'll take care of this in the next days
unless somebody objects or I find more issues.
The more complex problem "LVM over AoE" was not supported in jessie so
there's no reason to call this a blocker for stretch. It still was nice
to have it supported in Debian. Some day.
Christoph
signature.asc
Description: Digital signature
