Has anyone figured a good method for using jails following the method in section 15.6 of the Handbook while using zfs with quotas?
Right now each jailspace has its own zpool with a quota in /z/jail/${jail}. The mroot is just in the general pool without a quota also in /z/jail. I nullfs RO mount the mroot to /z/jail/mount/${jail} but the problem is the free disk space reported within the jail is that of mroot (which makes sense) and not that of the RW mount of the jail's jail space.
One seeming unavoidable, unappealing alternative is to have the mroot physically be in /z/jail/${jail} along with the jailspace and the jail's actual mount point. Doing a nullfs mount all within the same zfs pool should correctly report the available disk as expected however this makes updating less appealing.
Am I over engineering this? Did I even articulate my problem well enough that anyone has an idea what I'm trying to do?
Right now each jailspace has its own zpool with a quota in /z/jail/${jail}. The mroot is just in the general pool without a quota also in /z/jail. I nullfs RO mount the mroot to /z/jail/mount/${jail} but the problem is the free disk space reported within the jail is that of mroot (which makes sense) and not that of the RW mount of the jail's jail space.
One seeming unavoidable, unappealing alternative is to have the mroot physically be in /z/jail/${jail} along with the jailspace and the jail's actual mount point. Doing a nullfs mount all within the same zfs pool should correctly report the available disk as expected however this makes updating less appealing.
Am I over engineering this? Did I even articulate my problem well enough that anyone has an idea what I'm trying to do?