[BUG] Alternative default bastille_zfs_prefix

[MamoruDS]

Is your feature request related to a problem? Please describe.

The current default value of bastille_zfs_prefix has been changed(#711) to ${bastille_zfs_zpool}/bastille. Since all the scripts use ${bastille_zfs_zpool}/${bastille_zfs_prefix} as the dataset path prefix, this causes a duplication of bastille_zfs_zpool (which will be zroot/zroot/bastille after setup without any modification).

Describe the solution you'd like

To avoid this issue, manually change the config value of bastille_zfs_prefix to an arbitrary value (e.g., the previous bastille). Reverting to bastille as the default may provide a smoother experience.

Additional context

An example in bootstrap.sh shows the script using ${bastille_zfs_zpool}/${bastille_zfs_prefix} as the dataset path:

https://github.com/BastilleBSD/bastille/blame/cee6f20aa5c228e7e4791bc9d0fcdc412bab71ae/usr/local/share/bastille/bootstrap.sh#L150

[yaazkal]

Thanks for the comment and the workaround, I'm gonna take this

[tschettervictor]

There might need to be a variable added here.

Say someone wants to run Bastille on pool/dataset1/dataset2. The only way to make this work is by setting the zpool variable to pool/dataset1/dataset2 which is confusing because datasets are datasets and not pools.

Since the code has already been written to use only those two variables, one could just update the docs to let people know that zpool needs to be the full zfs path to the dataset where Bastille will reside. Otherwise it will always just default to “poolname/bastille”

the alternative of course is to submit #729 and add documentation telling people to set bastille_zfs_prefix to whichever dataset the want below the bastille_zfs_zpool with /bastille at the end.

[yaazkal]

@MamoruDS thanks for the report, PR #738 should fix it and it's now merged to the master branch.

@tschettervictor I guess bastille_zfs_prefix is where the dataset and path should reside in order to not create another variable that can end up touching a lot of files and maybe inserting some unwanted errors in the way. Are you willing to create a PR for the documentation?

[tschettervictor]

I'll give it a go, sure.

[tschettervictor]

#739

[yaazkal]

Closing this issue as the fix has been merged previously