From thomas.weissschuh at linutronix.de Tue Dec 10 11:17:09 2024 From: thomas.weissschuh at linutronix.de (Thomas =?utf-8?Q?Wei=C3=9Fschuh?=) Date: Tue, 10 Dec 2024 11:17:09 +0100 Subject: [elbe-users] elbe initvm ensure Fails In-Reply-To: References: Message-ID: <20241210111518-019a5632-18ae-4452-afb7-de06b31eedf2@linutronix.de> Hi Richard, sorry for the slow reply. On Fri, Sep 06, 2024 at 11:00:31PM -0700, Richard Cochran wrote: > Running on Debian stable host, I get: > > elbe initvm ensure > /home/richard/git/elbe/elbepack/initvm.py:239: KeyError: 'fdgroup' > > richard at hoboy:~/git/elbe$ gl -1 > 0a3e21954 (HEAD -> releases/v15.1, tag: v15.1, origin/releases/v15.1) release: v15.1 > > Any ideas? This should be fixed in release v15.2. Specifically it was fixed by commit 657e9bd10307 ("elbepack: initvm: handle stripped fdgroup arguments") Regards, Thomas From thomas.weissschuh at linutronix.de Tue Dec 10 14:44:24 2024 From: thomas.weissschuh at linutronix.de (Thomas =?utf-8?Q?Wei=C3=9Fschuh?=) Date: Tue, 10 Dec 2024 14:44:24 +0100 Subject: [elbe-users] Image with flat BTRFS Layout In-Reply-To: References: Message-ID: <20241210142209-395742ff-d0ce-4196-88d8-b1714017ebcb@linutronix.de> Hi Steffen, On Mon, Nov 18, 2024 at 12:31:13PM +0000, M?ller, Steffen - LT wrote: [..] > Now I want to implement a flat BTRFS design: > https://archive.kernel.org/oldwiki/btrfs.wiki.kernel.org/index.php/SysadminGuide.html#Subvolumes [..] > The image generation does work. But both subvolumes are empty and the > Debian image is installed on the subvolume=/. In addition the home > directory is empty too. > > Can someone help me? When copying filesystems to partitions currently only one, the last, entry defined in the XML is used. This may work by chance if a nested layout, where subvol names match their mountpoints, is used but for flat ones it breaks as you observe. Also mount options are not applied during building which would also be required to support the XML configuration you used. In my opinion it makes sense to get rid of this assumption. But coming up with something that can handle all usecases (also for example where one filesystem spans multiple block devices) will be some work. Regards, Thomas