EasyPBI cannot clean up .chroot dataset if snapshots exist (PCBSD 9.1)
|Reported by:||gja||Owned by:||kris|
I just used EasyPBI under PCBSD 9.1 (DVD x64 version) for the first time and stumbled across a quirk during "zfs destroy" after a build is successful.
I tweak my FreeBSD desktops to make zfs snapshots of all mounted zfs datasets every 15 minutes (and delete them every few hours). EasyPBI created a .chroot zfs dataset for its own use when I initiated "Build PBI". So now, while EasyPBI is doing its thing, my snapshot'ing script starts creating snapshots of the .chroot dataset every 15min.
When EasyPBI finishes building the local PBI, it trys to "zfs destroy" the .chroot dataset and fails. In particular, the log ends with:
[...] Cleaning /usr/pbi/tetex-base-amd64.chroot cannot destroy 'tank1/usr/pbi/tetex-base-amd64.chroot': filesystem has children use '-r' to destroy the following datasets: tank1/usr/pbi/tetex-base-amd64.chroot@121230_0845 tank1/usr/pbi/tetex-base-amd64.chroot@121230_0800 tank1/usr/pbi/tetex-base-amd64.chroot@121230_0900 tank1/usr/pbi/tetex-base-amd64.chroot@121230_0945 tank1/usr/pbi/tetex-base-amd64.chroot@121230_0815 tank1/usr/pbi/tetex-base-amd64.chroot@121230_0915 tank1/usr/pbi/tetex-base-amd64.chroot@121230_0930 tank1/usr/pbi/tetex-base-amd64.chroot@121230_0830
I suspect the solution is for EasyPBI to use "zfs destroy -r" rather than "zfs destroy", just to catch oddball cases like mine where the .chroot dataset ends up with children that really don't need to survive after the PBI build has succeeded.